Based on kernel version 4.9. Page generated on 2016-12-21 14:34 EST.
1 OCFS2 online file check 2 ----------------------- 3 4 This document will describe OCFS2 online file check feature. 5 6 Introduction 7 ============ 8 OCFS2 is often used in high-availability systems. However, OCFS2 usually 9 converts the filesystem to read-only when encounters an error. This may not be 10 necessary, since turning the filesystem read-only would affect other running 11 processes as well, decreasing availability. 12 Then, a mount option (errors=continue) is introduced, which would return the 13 -EIO errno to the calling process and terminate further processing so that the 14 filesystem is not corrupted further. The filesystem is not converted to 15 read-only, and the problematic file's inode number is reported in the kernel 16 log. The user can try to check/fix this file via online filecheck feature. 17 18 Scope 19 ===== 20 This effort is to check/fix small issues which may hinder day-to-day operations 21 of a cluster filesystem by turning the filesystem read-only. The scope of 22 checking/fixing is at the file level, initially for regular files and eventually 23 to all files (including system files) of the filesystem. 24 25 In case of directory to file links is incorrect, the directory inode is 26 reported as erroneous. 27 28 This feature is not suited for extravagant checks which involve dependency of 29 other components of the filesystem, such as but not limited to, checking if the 30 bits for file blocks in the allocation has been set. In case of such an error, 31 the offline fsck should/would be recommended. 32 33 Finally, such an operation/feature should not be automated lest the filesystem 34 may end up with more damage than before the repair attempt. So, this has to 35 be performed using user interaction and consent. 36 37 User interface 38 ============== 39 When there are errors in the OCFS2 filesystem, they are usually accompanied 40 by the inode number which caused the error. This inode number would be the 41 input to check/fix the file. 42 43 There is a sysfs directory for each OCFS2 file system mounting: 44 45 /sys/fs/ocfs2/<devname>/filecheck 46 47 Here, <devname> indicates the name of OCFS2 volume device which has been already 48 mounted. The file above would accept inode numbers. This could be used to 49 communicate with kernel space, tell which file(inode number) will be checked or 50 fixed. Currently, three operations are supported, which includes checking 51 inode, fixing inode and setting the size of result record history. 52 53 1. If you want to know what error exactly happened to <inode> before fixing, do 54 55 # echo "<inode>" > /sys/fs/ocfs2/<devname>/filecheck/check 56 # cat /sys/fs/ocfs2/<devname>/filecheck/check 57 58 The output is like this: 59 INO DONE ERROR 60 39502 1 GENERATION 61 62 <INO> lists the inode numbers. 63 <DONE> indicates whether the operation has been finished. 64 <ERROR> says what kind of errors was found. For the detailed error numbers, 65 please refer to the file linux/fs/ocfs2/filecheck.h. 66 67 2. If you determine to fix this inode, do 68 69 # echo "<inode>" > /sys/fs/ocfs2/<devname>/filecheck/fix 70 # cat /sys/fs/ocfs2/<devname>/filecheck/fix 71 72 The output is like this: 73 INO DONE ERROR 74 39502 1 SUCCESS 75 76 This time, the <ERROR> column indicates whether this fix is successful or not. 77 78 3. The record cache is used to store the history of check/fix results. It's 79 default size is 10, and can be adjust between the range of 10 ~ 100. You can 80 adjust the size like this: 81 82 # echo "<size>" > /sys/fs/ocfs2/<devname>/filecheck/set 83 84 Fixing stuff 85 ============ 86 On receiving the inode, the filesystem would read the inode and the 87 file metadata. In case of errors, the filesystem would fix the errors 88 and report the problems it fixed in the kernel log. As a precautionary measure, 89 the inode must first be checked for errors before performing a final fix. 90 91 The inode and the result history will be maintained temporarily in a 92 small linked list buffer which would contain the last (N) inodes 93 fixed/checked, the detailed errors which were fixed/checked are printed in the 94 kernel log.