+ SnapRAID
+ Frequently Asked Questions
+Common questions about configuration, recovery, and usage.
+ + + + + ++General +
++What is SnapRAID? +
++SnapRAID is an application that creates a partial backup of your disk array. +If some disks in your array fail, even if they are completely broken, you can recover their contents. +It provides only a partial backup because it does not allow recovery from a failure of the entire array, +but only if the number of failed disks is below a predefined limit. +
++How is it different from RAID? +
++RAID solutions mirror data and propagate changes immediately to the parity data. +In this sense, RAID systems are not backups because they do not allow restoration of an earlier state of the array. +
++SnapRAID, however, is more akin to a backup solution, allowing you to restore the state of the last backup. +For example, with SnapRAID, if you accidentally delete a file, you can easily recover it. +
++Other advantages over traditional RAID5/6 solutions include: +
+-
+
- +All your data is hashed to ensure data integrity and prevent silent corruption. + +
- +If too many disks fail to allow recovery, you lose data only on the failed disks; data on other disks remains safe. + +
- +If you accidentally delete files on a disk, you can recover them. + +
- +You can start with already filled disks. + +
- +Disks can have different sizes. + +
- +You can add disks at any time. + +
- +It does not lock in your data. You can stop using SnapRAID at any time without needing to reformat or move data. + +
- +To access a file, only a single disk needs to spin, saving power and reducing noise. + +
- +You can have up to six parity levels, compared to one for RAID5 and two for RAID6. + +
+RAID has advantages, such as better speed due to data striping. +The best choice depends on your specific use case. +
++How is it different from ZFS/Btrfs snapshots and NTFS Shadow Copy? +
++SnapRAID creates a file-system snapshot conceptually similar to other snapshot solutions. +However, its primary purpose is to recover a disk to its saved state after a complete disk failure. +
++SnapRAID is more akin to the RAID-Z/RAID functionality of ZFS/Btrfs. +Compared to them, SnapRAID offers the following advantages: +
+-
+
- +If too many disks fail to allow recovery, you lose data only on the failed disks; data on other disks remains safe. + +
- +You can start with already filled disks. + +
- +You can add disks at any time (not possible with ZFS, but possible with Btrfs). + +
- +It does not lock in your data. You can stop using SnapRAID at any time without needing to reformat or move data. + +
- +To access a file, only a single disk needs to spin, saving power and reducing noise. + +
- +You can have up to six parity levels, compared to three for ZFS RAID-Z and two for Btrfs RAID-5/6. + +
+ZFS and Btrfs have advantages, such as being real-time solutions. +The best choice depends on your specific use case. +
++How is it different from other backup solutions? +
++With a typical backup solution, you need as much space as the entire disk array for the backup. +With SnapRAID, you need only one additional parity disk to recover from the most common case of a single disk failure. +Adding more parity disks allows recovery from additional disk failures. +
++Note that only a full backup allows recovery from a complete failure of the disk array, +making it the preferred solution if you can afford it. +SnapRAID is a good option when a full backup is not feasible. +
++What are the advantages of SnapRAID compared to similar solutions like unRAID, and Windows Storage Spaces? +
++Visit the + +Compare + +page to see a comparison with other similar solutions. +
++A key advantage of SnapRAID is its + +integrity and silent error management + +, which is at the same level as (or better than) +ZFS and Btrfs file systems. In contrast, unRAID and Windows Storage Spaces lack integrity checks. +
++How much does it cost? +
++Nothing. SnapRAID is a + +free software + +application released under the GPLv3 license. +
++How can I get support? +
++Visit the + +Support + +page. +
++Does SnapRAID have a GUI? +
++No, SnapRAID is a command-line application. +However, there is an external GUI called + +Elucidate + +, +and a plugin for + +OpenMediaVault + +. +
++Does SnapRAID provide encryption, virtual views/storage pooling, SMART monitoring, or power control? +
++SnapRAID provides a basic virtual views/storage pooling solution but is also compatible with other pooling solutions. +
++For encryption, SMART monitoring, power control, and data recovery, you can use other tools. +Here are some suggestions: +
+| + | ++Linux + | ++Windows + | ++Mac OS X + | +
|---|---|---|---|
| +Encryption + | +
+
+ dm-crypt/LUKS
+
+ +( + + tutorial + +) + |
++ + VeraCrypt + + | ++ | +
| +Virtual views/storage pooling + | ++ + mergerfs + +( + + tutorial + +), + + Greyhole + + | ++ + DrivePool ($) + + | ++ | +
| +SMART monitoring + | ++ + smartmontools + +, + + GSmartControl + + | ++ + smartmontools + +, + + GSmartControl + +, + + CrystalDiskInfo + + | ++ + smartmontools + +, + + GSmartControl + +, + + SMARTReporter + + | +
| +Power control + | ++ + hdparm + + | ++ + hdparm + + | ++ | +
| +Data recovery + | ++ + ddrescue + +, + + safecopy + +, + + dd_rescue + + | ++ | ++ | +
+Is SnapRAID suitable for backing up the home/My Documents directory? +
++No. If you have a limited amount of data that can be easily saved to an external USB drive +or an online backup service, using SnapRAID is unnecessary. +SnapRAID is best suited for backing up a + +very large + +amount of data, +such as many terabytes, where a full backup copy is not feasible. +
++Is SnapRAID suitable for backing up the boot OS disk? +
++No. For the disk used to boot your machine and containing the OS files, +it’s better to use another solution, such as RAID1 or a mirror copy. +SnapRAID requires an operating system to run and cannot recover the OS if it is missing. +
++Can SnapRAID hash files without parity computation? +
++No, SnapRAID cannot be used solely to hash files. Its hashing is designed to work in combination with parity computation +and is not optimized for hashing alone. +
++For a list of file verification tools, see: + +Wikipedia: Comparison of file verification software + +. +
++For a setup example in Ubuntu 14.04, see + +Zack's SnapRAID tutorial + +. +
++Setup +
++Do I need one or more parity disks? +
++As a rule of thumb, use one parity disk (RAID5) for up to four data disks, +and then add one parity disk for each group of seven data disks, as shown in the table: +
+| +Parities + | ++Data disks + | +
|---|---|
| +1/Single Parity/RAID5 + | ++2–4 + | +
| +2/Double Parity/RAID6 + | ++5–14 + | +
| +3/Triple Parity + | ++15–21 + | +
| +4/Quad Parity + | ++22–28 + | +
| +5/Penta Parity + | ++29–35 + | +
| +6/Hexa Parity + | ++36–42 + | +
+This table accounts for the fact that more parity levels help recover from an unsynced array +and that disk failures may be correlated if all disks are in the same environment. +
++Multiple failures can occur. The worst case reported in SnapRAID’s history +is a + +four-disk failure, successfully recovered using four parities + +. +
++For more information on parity levels and disk failure rates, see: + +Why RAID 5 stops working in 2009 + +, + +Why RAID 6 stops working in 2019 + +, + +Does RAID 6 stop working in 2019? + +, + +Triple-Parity RAID and Beyond + +, + +How long do disk drives last? + +. +
++Which file system is recommended for SnapRAID? +
++For data disks, consider these suggestions: +
+| +Data + | ++Result + | +
|---|---|
| +Any + | ++OK. You can generally use any file system you prefer, with some exceptions noted below. + | +
| +Btrfs + | ++OK. Ensure SnapRAID is built with the libblkid library for UUID support. +You will be warned at each run if this is not the case. +Note that multiple Btrfs snapshots are not supported. + | +
| +ZFS + | ++Partial. UUID support is unavailable, but core SnapRAID functionalities remain usable. +Note that multiple ZFS snapshots are not supported. + | +
| +FAT + | ++Uncertain. There are + + reports + +of duplicate inode cases. NTFS is significantly faster and recommended instead. + | +
| +ReFS + | ++No. ReFS uses + + 128-bit inodes + +, +which SnapRAID does not yet support. + | +
+For parity disks, consider these suggestions: +
+| +Parity + | ++Result + | +
|---|---|
| +ext4 + | ++OK if disks are not larger than 16 TB. The parity is stored in a single large file, and ext4 has a 16 TB file size limit. + | +
| +XFS + | ++OK + | +
| +Btrfs + | ++OK. Ensure SnapRAID is built with the libblkid library for UUID support. +You will be warned at each run if this is not the case. + | +
| +JFS + | ++No. It does not support the + + fallocate() + +command needed to allocate parity files. +It also does not optimally implement the + + posix_fadvise() + +command. + | +
| +NTFS + | ++OK in Windows. Not supported in Linux, as it does not support the + + fallocate() + +command needed to allocate parity files. +You can use it if the parity disk contains only the parity file. + | +
| +FAT + | ++No. Files larger than 4 GB cannot be stored. It is also not journaled. Use NTFS instead. + | +
| +HFS+ + | ++OK + | +
+In Linux, to maximize space for the parity file system, format it with the + +-m 0 -T largefile4 + +options, like: +
+mkfs.ext4 -m 0 -T largefile4 DEVICE+
+On an 8 TB disk, this can save about 400 GB. It is also expected to be as fast as, or faster than, the default. +
++Which hardware configuration is recommended? +
++For optimal performance, connect all disks via SATA, not USB. +
++See + +Raj's Prototype Builds + +for examples of hardware setups. +
++For choosing hard drives, refer to + +Backblaze's What Hard Drive Should I Buy? + +and + +Backblaze's Enterprise Drives: Fact or Fiction? + +. +
++Is a multi-core CPU recommended? +
++No. SnapRAID is typically not CPU-bound; the limiting factor is disk read performance. +A single-core CPU is usually sufficient. +
++For example, in my system, I achieve a combined read speed of 400 MiB/s, +but SnapRAID can hash at 12,000 MiB/s on a single core. +
++Is a 64-bit operating system recommended? +
++Yes. SnapRAID performs better on a 64-bit operating system, as it can access more than 4 GiB of memory +and use faster hashing and parity algorithms. +However, for less than 30 TB of data, it performs well on a 32-bit operating system too. +
++How much memory does SnapRAID require to run? +
++Approximately, SnapRAID requires 1 GiB of memory for every 16 TB of data. +
++If you have insufficient memory, the recommended solution is to install more memory. +
++Alternatively, increase the blocksize option in the configuration file. +Memory usage is inversely proportional to the selected block size. +For example, using "blocksize 512" instead of the default 256 halves memory usage. +However, a larger block size increases wasted space in parity files. +
++Running with memory swapped to disk is not recommended, as it significantly slows performance. +
++Should I check the SMART attributes of my disks? +
++Yes. Approximately 20% of disks fail within four years, as +reported by + +Backblaze's How long do disk drives last? + +. +Monitoring SMART attributes can help anticipate failures and prevent data loss. +
++If any of the SMART attributes + +Reallocated_Sector_Ct (5) + +, + +Reported_Uncorrect (187) + +, + +Command_Timeout (188) + +, + +Current_Pending_Sector (197) + +, +or + +Offline_Uncorrectable (198) + +are non-zero, replace the disk, as recommended by + +Backblaze's Hard Drive SMART Stats + +. +
++See also + +Google's Failure trends in a large disk drive population + +, +which found: +
+After the first reallocation, drives are +over 14 times more likely to fail within +60 days than drives without reallocation +counts.+
After the first probational event, drives +are 16 times more likely to fail within +60 days than drives with zero probational +counts.+
After the first offline reallocation, drives have over +21 times higher chances of failure within 60 days than +drives without offline reallocations.+
+These findings recommend replacing a disk after the first error in the SMART attributes + +Reallocated_Sector_Ct (5) + +, + +Current_Pending_Sector (197) + +, or + +Offline_Uncorrectable (198) + +. +
++Is it possible to use SnapRAID in a dual-boot configuration, such as Windows and Linux? +
++Yes, SnapRAID can be used in a dual-boot configuration, regardless of the OS combination. +
++It is recommended to run SnapRAID from only one OS. +While it may work if run from both OSes, this scenario is not thoroughly tested and risks issues +due to different SnapRAID versions, configurations, or content files. +
++Configuration +
++What is the SnapRAID 'content' file? +
++The content file stores the list of all files in your array, including their checksums, timestamps, and other necessary information. +
++This file will be approximately a few GiB in size, depending on the array’s size. +For 10 TB of data, the content file will be about 500 MiB. +
++What are the SnapRAID 'parity' and 'N-parity' files? +
++These files store the parity data used for recovery. +
++Their size grows to match the largest amount of data stored on any single disk in the array. +
++How do I configure SAMBA to share the pool directory? +
++To configure SAMBA to share the pool directory (e.g., /pool), add the following options to +your /etc/samba/smb.conf: +
++# In the global section of smb.conf +unix extensions = no + +# In the share section of smb.conf +[pool] +comment = Pool +path = /pool +read only = yes +guest ok = yes +wide links = yes ++
+How can I use more disks than available drive letters in Windows? +
++Use + +Volume Mount Points + +to mount a disk in a directory without +using a drive letter, similar to Linux. +
++Use +
++What are the most important steps to prevent data loss? +
++To minimize the risk of data loss, follow these steps: +
+-
+
- +Before running the first "sync", test your RAM with a program +like + +MemTest86 + +. +Faulty RAM is the most common cause of data loss with SnapRAID. + +
- +Run the "sync" command frequently, from once a day to once a week. + +
- +Run the "scrub" command weekly. + +
- +Use + +smartmontools + +or a similar utility +to monitor SMART attributes. +Replace a disk at the first sign of + +Reallocated_Sector_Ct + +, + +Current_Pending_Sector + +, or + +Offline_Uncorrectable + +, +even if it still functions. + +
+How can I add a data disk to an existing array? +
++To add a new data disk to the array, add the new "disk" option in the configuration file, +then run a "sync" command: +
++snapraid sync ++
+If the disk is empty, the "sync" command will complete immediately. +
++How can I remove a data disk from an existing array? +
++To remove a data disk from the array, follow these steps: +
+-
+
- +Change the related "disk" option in the configuration file to point to an empty directory. + +
- +Remove any "content" option pointing to that disk from the configuration file. + +
-
+Run a "sync" command with the "-E, --force-empty" option:
+
+snapraid sync -E +
+The "-E" option allows SnapRAID to proceed even when detecting an empty disk. +
+ - +After the "sync" command completes, remove the "disk" option from the configuration file. + +
+The array will then have no reference to the removed disk. +
++How can I add or remove a parity disk in an existing array? +
++To add a new parity level, add the appropriate "N-parity" option in the configuration file, then run the "sync" command +with the "-F, --force-full" option: +
++snapraid -F sync ++
+The "-F" option instructs SnapRAID to recompute the full parity. +
++During this process, your existing parity remains unmodified, ensuring continued protection +(note that this applies only from version 11.0). +
++To remove a parity level, remove the highest "N-parity" option from the configuration file and delete the parity file. +
++Note that after removing a parity file, it cannot be reused, as it becomes outdated after the next "sync" command. +
++How can I verify a single data disk? +
++The "scrub" command verifies the entire array and cannot be limited to a single data disk. +To verify only one data disk, use the "check" audit command with the disk specified: +
++snapraid check -a -d DISK_NAME ++
+This command verifies the checksums of all files on the specified disk. +Other data and parity disks are not read or verified. +
++How can I replace a data disk? +
++If the disk is lost, refer to the recovery section in the manual. +Otherwise, copy all files to the new disk, maintaining the same directory structure and names. +
++In Linux, use the command: +
++cp -av /from_dir/. /to_dir ++
+WARNING: If using 'rsync' or 'mc' commands, ensure they are recent versions. +'rsync' must be at least version 3.1.0 (2013), and 'mc' at least version 4.8.19 (2017). +Older versions may not copy sub-second timestamps. +
++In Windows, use the command: +
++robocopy F:\from_dir T:\to_dir /e /copyall ++
+Update the SnapRAID configuration to point to the new mount point, then run a "diff" command to ensure everything is correct: +
++snapraid diff ++
+If "diff" reports "added" or "removed" files, you may have copied files with a different directory structure. +Correct this until "diff" reports only "equal", "restored", or "copied" files. +
++To verify that files were copied correctly, run a "check" audit command on the replaced disk to verify checksums: +
++snapraid check -a -d DISK_NAME ++
+Finally, run "sync" to update the SnapRAID state. This command will complete almost immediately: +
++snapraid sync ++
+How can I replace a parity disk? +
++If the disk is still accessible, ensure the array is synced by running: +
++snapraid sync ++
+Copy the parity file to the new disk and update the parity location in the configuration file. +
++Run "sync" again to make SnapRAID recognize the new disk. This command will complete almost immediately: +
++snapraid sync ++
+In Linux, a good copy command is + +ddrescue + +. For example: +
++ddrescue /OLD_DISK/parity /NEW_DISK/parity /tmp/copy.log ++
+This copies the parity file from the old to the new location, continuing despite errors. +To retry problematic sections, repeat the same command. +
++A partial parity file is still useful, as it allows recovery using the valid parts. +
++If the parity file is partially damaged or lost, use the "fix" command with the -d option to recreate it, targeting only the affected parity: +
++snapraid fix -d PARITY_NAME ++
+How can I safely move files from one disk to another? +
++To safely move files between data disks in the array, follow these steps: +
+-
+
- +Copy the files to the new location. + +
- +Run a 'diff' command to verify that the copied files are identified as 'copy'. + +
- +Run a 'sync' command. + +
- +Delete the files from the original location. + +
- +Run another 'sync' command. + +
+This process ensures that, even if disks fail during the sync process, +you can recover all data. It also verifies that the hashes of moved files match the originals. +
++Can I defragment data and parity disks? +
++Yes, defragmentation does not affect SnapRAID. +
++How can I make SnapRAID less resource-intensive for other processes? +
++In Linux, use the + +ionice + +command with idle priority, like: +
++ionice -c 3 snapraid sync ++
+How can I upgrade to a newer version of SnapRAID? +
++To upgrade, replace the old SnapRAID executable with the new one. +The new version will use your existing configuration, content, and parity files. +
++Occasionally, the content file format changes, but newer SnapRAID versions can read all older formats. +The content file is updated to the new format the next time it is written, if necessary. +
++Downgrading to an older SnapRAID version is generally not possible. +If the old version cannot read the new content file, you will receive an appropriate message. +
++Recover +
++How can I undelete a recently deleted file? +
++Run the "fix" command with a filter for the specified file, like: +
++snapraid fix -f my_just_deleted_file ++
+To undelete a directory, use: +
++snapraid fix -f my_just_deleted_dir/ ++
+To undelete all missing files, use: +
++snapraid fix -m ++
+What happens if I add files and a disk fails before updating parity with "sync"? +
++Files added to the failed disk after the last "sync" command are lost. +All other data is safe and can be recovered. +
++What happens if I delete files and a disk fails before updating parity with "sync"? +
++In the worst case, files deleted or modified on non-failed disks may prevent recovery of an equivalent amount of data on the failed disk. +For example, deleting 10 GB of data may prevent recovery of 10 GB from the failed disk. +The exact data loss depends on how much the deleted and failed data overlap in the parity. +
++Using two parity disks significantly improves recovery chances. +
++To ensure safety, avoid deleting files directly. Instead, move them to another directory on the same disk, +outside the SnapRAID-checked directory tree. Run the "sync" command, then delete the files. +If a disk fails during the "sync", run the "fix" command, specifying the location of the deleted files: +
++snapraid fix -i dir_of_deleted_files ++
+What happens if a disk fails during a "sync"? +
++You can still recover data. In the worst case, you can recover as much data as if the disk failed before the "sync". +If the "sync" process has progressed, SnapRAID uses the partially synced data to recover more. +Using the "autosave" configuration option saves intermediate content files during the sync process, improving recovery. +
++What are "data errors" in "sync", and how do I fix them? +
++Syncing... +309505: Data error for file d4/data/index.db at position 40960 +WARNING! Unexpected data error in a data disk! The block is now marked as bad! +Try with 'snapraid -e fix' to recover! ++
+Data errors occur when SnapRAID detects that a file’s content has changed without an updated write timestamp. +This should not happen and may indicate a silent disk error or a file system issue, such as an improper shutdown. +
++If a data error is found, the "sync" process continues, marking erroneous blocks as bad. +These can be fixed with the "-e fix" command, which typically takes only a few seconds: +
++snapraid -e fix ++
+If errors occur in multiple files, check for hardware issues. +Test your PC’s memory with a tool like + +MemTest86 + +. +Also, check disk cabling and CPU temperature. +
++Another possibility is that a program modified files without updating their timestamps. +In Linux, this is rare, except for memory-mapped files not yet closed. +In Windows, timestamps are guaranteed to update only after closing a file. +See + +MSDN WriteFile() + +: +
++When writing to a file, the last write time is not fully updated until +all handles used for writing have been closed. Therefore, to ensure an +accurate last write time, close the file handle immediately after writing +to the file. ++
+For memory-mapped files in Windows, the program must manually update the timestamp, +as Windows does not do so automatically. +See + +MSDN CreateFileMapping() + +: +
++When modifying a file through a mapped view, the last modification timestamp +may not be updated automatically. If required, the caller should use SetFileTime +to set the timestamp. ++
+To avoid these issues in Windows, run "sync" when no programs are accessing files in the array. +
++How can I store all messages from a check or fix command in a file? +
++Use the "-l" option to generate a log file: +
++snapraid check -l mylog.log ++
+These log files can be large, potentially several GB. If too large for your editor, +use + +HxD + +. +
++How can I recover the snapraid.conf file? +
++Use the "-C, --gen-conf" command to generate a template configuration file +compatible with your content file. +Manual adjustments are needed to set disk mount points, but comments in the file provide guidance: +
++snapraid -C snapraid.content > snapraid.conf.new ++
+Tech +
++If the SnapRAID "check" command reports everything is OK, is it really OK? +
++Yes. SnapRAID verifies all files with a 128-bit checksum. +If it reports they are correct, they are. +
++What integrity checksum does SnapRAID use? +
++SnapRAID uses + +Murmur3 + +and + +SpookyHash + +128-bit hashes. +Murmur3 is the default on 32-bit platforms, and SpookyHash is the default on 64-bit platforms. +
++Why are integrity checksums important? +
++Integrity checksums detect silent errors when a hard drive returns corrupt data without reporting an error. +
++For an introduction to this issue, see: + +Data corruption is worse than you know + +. +
++For a detailed analysis, see: + +An Analysis of Data Corruption in the Storage Stack + +: +
++A significant number (8% on average) of corruptions +are detected during RAID reconstruction, creating the +possibility of data loss. In this case, protection against +double disk failures is necessary to prevent data loss. ++
+How can SnapRAID recover if a disk fails during a 'sync'? +
++Before starting a sync, SnapRAID saves a special content file describing both the pre- and post-sync states. +This allows SnapRAID to attempt two recovery strategies. +
++For each block to recover, it first assumes the parity was correctly computed, as if the "sync" completed successfully. +If this fails, it assumes the parity was not yet updated, ignoring any new files added to the array. +
++Using double parity improves recovery when there are overlapping changes in the array. +
++Can I access data disks during a 'sync'? +
++Yes, you can safely read from data disks during a 'sync'. +
++You can also write, and if this interferes with the 'sync', +the process continues, skipping the written part. +
++This may affect potential recovery, but it is equivalent to writing just after the 'sync' completes. +
++Does SnapRAID retry reads or writes if a disk fails? +
++No, SnapRAID does not retry disk operations. It reports errors and attempts to continue. +After 100 errors, it stops execution. +
++In such cases, it’s better to allow manual data recovery before the disk fails completely. +Automatic retries could further damage the disk. +To retry, repeat the command manually. +
++You can adjust the 100-error limit using the "-L, --error-limit" option. +
++For "fix" and "check" commands, you can resume from the point of interruption using the "-s, --start" option, specifying the block number that caused the error. +
++For the "sync" command, the latest state is saved before termination, so restarting it will continue from where it stopped. +
++What is the snapraid.content.lock file? +
++This file is used by SnapRAID to check if another instance is running. +It is created where the first content file is stored. +
++What does the notice about zero sub-second timestamps mean? +
++In the 'status' command, you may see: +
++You have XXXXX files with zero sub-second timestamp. +Run the 'touch' command to set it to a non-zero value. ++
+This indicates that some files have timestamps rounded to the nearest second, +even though your file system supports sub-second precision. +
++While not a problem, this may limit SnapRAID’s ability to detect moved or copied files between disks. +
++You can set sub-second timestamps to a random value using the 'touch' command, which does not modify the second part of the timestamp. +
++Performance +
++What is the optimal hardware configuration for the best performance? +
++The most critical factor is using (E)SATA connections for all disks to avoid controller slowdowns. +Avoid USB disks. +
++The second factor is using disks with similar performance, as the array’s read/write speed is limited by the slowest disk. +After a sync, SnapRAID displays statistics with wait times for each disk, helping identify bottlenecks. +
++Finally, a fast CPU can improve performance with many disks. +Check CPU usage during sync operations to determine if it’s a limiting factor. +
++What is the optimal block size for the best performance? +
++The default block size is typically optimal. +
++Increasing it may not improve performance and could reduce it in some cases. +
++What is the optimal Linux read-ahead size for the best performance? +
++The default Linux read-ahead size of 128 KiB is optimal for SnapRAID’s default block size of 256 KiB. +If using a larger block size, set the read-ahead size to at least half the SnapRAID block size. +
++For example, with a SnapRAID block size of 512 KiB, configure a read-ahead size of 256 KiB for sdX using: +
++echo 256 > /sys/block/sdX/queue/read_ahead_kb ++
+For a SnapRAID block size smaller than 256 KiB, use the default read-ahead size. +
++How fast are the hash and parity computations? +
++Murmur3 hash is very fast on 32-bit systems, and SpookyHash is even faster on 64-bit systems. +Parity computation is efficient up to double parity, even without SSE support. +Beyond double parity, a CPU with SSSE3 support is recommended. +For low-end CPUs without SSE, a special, incompatible triple-parity option is available. +See 'z-parity' in the manual for details. +
++The following results are from a Intel Core i7-10700 CPU @ 2.90GHz. +Check your values with the -T option: +
++root@redstar:/root# snapraid -T +snapraid v13.0 by Andrea Mazzoleni, http://www.snapraid.it +Compiler gcc 15.2.0 +CPU GenuineIntel, family 6, model 165, flags sse2 ssse3 crc32 avx2 +Memory is little-endian 64-bit +Support nanosecond timestamps with futimens() + +Speed test using 8 data buffers of 262144 bytes, for a total of 2048 KiB. +Memory blocks have a displacement of 1792 bytes to improve cache performance. +The reported values are the aggregate bandwidth of all data blocks in MB/s, +not counting parity blocks. + +Memory write speed using the C memset() function: +memset 61914 + +CRC used to check the content file integrity: +table 1866 +intel 12741 + +Hash used to check the data blocks integrity: + best murmur3 spooky2 metro +hash spooky2 6285 17813 20684 + +RAID functions used for computing the parity with 'sync': + best int8 int32 int64 sse2 sse2e ssse3 ssse3e avx2 avx2e +gen1 avx2 24995 46562 62683 77069 +gen2 avx2 8243 15879 25788 27292 44907 +genz avx2e 5509 10588 14485 13920 25175 +gen3 avx2e 1190 13201 14307 27268 +gen4 avx2e 914 10061 11006 21328 +gen5 avx2e 746 7970 8828 17040 +gen6 avx2e 630 6627 7516 14584 + +RAID functions used for recovering with 'fix': + best int8 ssse3 avx2 +rec1 avx2 1626 3949 4331 +rec2 avx2 709 1738 2231 +rec3 avx2 170 922 1414 +rec4 avx2 110 614 981 +rec5 avx2 79 426 705 +rec6 avx2 58 316 539 ++
+Known Problems +
++Why does my machine reboot or crash while SnapRAID is running? +
++SnapRAID pushes the system to extreme conditions, which may expose latent hardware or driver issues, +resulting in a system crash or reboot. +
++Although the issue occurs only when SnapRAID is running, it is undoubtedly a hardware or driver problem, +as SnapRAID is a standard user application that cannot crash or reboot the system. +
++Check the following: +
-
+
- +Test system memory with + + MemTest86 + +. + +
- +Verify hard drive cabling. + +
- +Ensure the power supply can support all hard drives spinning simultaneously. + +
- +Check that the system temperature is not too high. + +
- +Update storage-related system drivers. + +
- +Review system logs or Event Viewer for clues about the issue. + +
+Why does SnapRAID abort without an error message in Linux? +
++This is likely due to insufficient free memory, causing the Linux out-of-memory (OOM) killer +to terminate SnapRAID for using too much memory. +Check available memory with: +
++free -m ++
+If memory is low, install more memory in the machine. +
++Why do I get the error 'Failed to grow parity file 'xxx' to size xxx due to lack of space' during 'sync'? +
++This error indicates that SnapRAID needs to expand the parity file to a size that exceeds the available space on the parity disk. +
++Ensure no other data is stored on the parity disk, reserving it exclusively for the parity file. +Alternatively, move files mentioned in the output to another data disk to reduce the required parity size. +
++In Linux, you can reformat the parity disk to increase available space using: +
++mkfs.ext4 -m 0 -T largefile4 DEVICE ++
+The '-T largefile4' and '-m 0' options provide more space for the parity file. +
++Why do I get the error 'Internal inode xxx inconsistency...'? +
++This error occurs when SnapRAID detects two different files using the same inode, +a condition that should theoretically never happen but can occur in specific cases. +
++Ensure the mentioned files were not modified while SnapRAID was running. +Renaming, deleting, or recreating files during SnapRAID operation can cause inode reuse, triggering this error. +
++If using a FAT file system in Windows, convert to NTFS. +Although inodes (FileIndex) are theoretically + +unique in FAT + +, +this issue has been observed in practice. +
++Why do I get the error 'Data change at file...' with .xls files? +
++Excel modifies .xls files upon opening, even without saving, to store the name of the last user who accessed the file. +This allows it to warn if another user tries to open the file simultaneously. +Excel restores the original timestamp, but SnapRAID detects the modification, reporting a data error: +
++Data change at file '/xxx/xxx.xls' at position '0' +WARNING! Unexpected data modification of a file without parity! ++
+To work around this, see + +How to prevent Excel from modifying the file on exit? + +, +or switch to the .xlsx format. +
++Why are VeraCrypt containers never saved? +
++VeraCrypt (a fork of TrueCrypt) has a default option, + +Preserve modification timestamp of file containers + +, +that prevents SnapRAID and other backup programs from detecting changes to file containers. +Disable this option in VeraCrypt. +
+
+ 
+ 
+ 
+ 
+