Score:2

After filesystem reached 100% storage capacity now set to read-only, how to reset to read-write mode?

lu flag

Yesterday our server (Ubuntu 18.04) reached 100% storage capacity enter image description here and set one of our filesystems to read-only mode, see: /dev/md3 / ext4 ro,relatime,errors=remount-ro,data=ordered 0 0. I've tried several solutions from other answers on serverfault, but none seem to fit my case.

For example, I've tried to execute the following command: sudo mount -o remount,rw /dev/md3 /, but that results in the message: mount: /: cannot remount /dev/md3 read-write, is write-protected.

How can I solve this problem to make the filesystem read-write again?

Thanks!

Update with debug info:

mdadm --detail /dev/md3
/dev/md3:
           Version : 0.90
     Creation Time : Fri Nov 10 10:07:34 2017
        Raid Level : raid1
        Array Size : 20478912 (19.53 GiB 20.97 GB)
     Used Dev Size : 20478912 (19.53 GiB 20.97 GB)
      Raid Devices : 2
     Total Devices : 2
   Preferred Minor : 3
       Persistence : Superblock is persistent

       Update Time : Sat Sep 18 09:15:35 2021
             State : clean
    Active Devices : 2
   Working Devices : 2
    Failed Devices : 0
     Spare Devices : 0

Consistency Policy : unknown

              UUID : 4b632ac4:ae1a7c2b:a4d2adc2:26fd5302
            Events : 0.861

    Number   Major   Minor   RaidDevice State
       0       8        3        0      active sync   /dev/sda3
       1       8       19        1      active sync   /dev/sdb3

And using dmesg:

dmesg | grep "md3"
[67448453.830094] EXT4-fs error (device md3): ext4_remount:4840: Abort forced by user

Executed tune2fs:

tune2fs -l /dev/md3
tune2fs 1.44.1 (24-Mar-2018)
Filesystem volume name:   /
Last mounted on:          /
Filesystem UUID:          d1a985c4-8c5e-4034-93e0-629b8e65f161
Filesystem magic number:  0xEF53
Filesystem revision #:    1 (dynamic)
Filesystem features:      has_journal ext_attr resize_inode dir_index filetype needs_recovery extent flex_bg sparse_super large_file huge_file uninit_bg dir_nlink extra_isize
Filesystem flags:         signed_directory_hash
Default mount options:    user_xattr acl
Filesystem state:         clean with errors
Errors behavior:          Continue
Filesystem OS type:       Linux
Inode count:              1281120
Block count:              5119728
Reserved block count:     255986
Free blocks:              445848
Free inodes:              1001361
First block:              0
Block size:               4096
Fragment size:            4096
Reserved GDT blocks:      1022
Blocks per group:         32768
Fragments per group:      32768
Inodes per group:         8160
Inode blocks per group:   510
Flex block group size:    16
Filesystem created:       Fri Nov 10 10:07:39 2017
Last mount time:          Tue Jul 30 17:51:41 2019
Last write time:          Thu Sep 16 20:06:05 2021
Mount count:              7
Maximum mount count:      -1
Last checked:             Fri Nov 10 10:07:39 2017
Check interval:           0 (<none>)
Lifetime writes:          4013 GB
Reserved blocks uid:      0 (user root)
Reserved blocks gid:      0 (group root)
First inode:              11
Inode size:           256
Required extra isize:     28
Desired extra isize:      28
Journal inode:            8
First orphan inode:       663035
Default directory hash:   half_md4
Directory Hash Seed:      ae316af1-086d-470f-af27-0c10ca25f3c8
Journal backup:           inode blocks
FS Error count:           8
First error time:         Thu Sep 16 20:06:04 2021
First error function:     ext4_lookup
First error line #:       1607
First error inode #:      930317
First error block #:      0
Last error time:          Sat Sep 18 09:15:35 2021
Last error function:      ext4_remount
Last error line #:        4840
Last error inode #:       685456
Last error block #:       0

Debug info using e2fsck -n /dev/md3:

e2fsck -n /dev/md3
e2fsck 1.44.1 (24-Mar-2018)
Warning: skipping journal recovery because doing a read-only filesystem check.
/ contains a file system with errors, check forced.
Pass 1: Checking inodes, blocks, and sizes
Inodes that were part of a corrupted orphan linked list found.  Fix? no

Inode 101 was part of the orphaned inode list.  IGNORED.
Inode 117 was part of the orphaned inode list.  IGNORED.
Inode 292 was part of the orphaned inode list.  IGNORED.
Inode 460 was part of the orphaned inode list.  IGNORED.
Inode 465 was part of the orphaned inode list.  IGNORED.
Inode 471 was part of the orphaned inode list.  IGNORED.
Inode 487 was part of the orphaned inode list.  IGNORED.
Inode 529 was part of the orphaned inode list.  IGNORED.
Inode 562 was part of the orphaned inode list.  IGNORED.
Inode 564 was part of the orphaned inode list.  IGNORED.
Inode 707 was part of the orphaned inode list.  IGNORED.
Inode 723 was part of the orphaned inode list.  IGNORED.
Inode 918 was part of the orphaned inode list.  IGNORED.
...
Deleted inode 402614 has zero dtime.  Fix? no
...
Inode 783370, end of extent exceeds allowed value
    (logical block 1024, physical block 3068928, len 76)
Clear? no

Inode 783370, i_blocks is 8784, should be 8200.  Fix? no

Inode 783470, end of extent exceeds allowed value
    (logical block 2708, physical block 1322783, len 193)
Clear? no

Inode 783470, i_blocks is 23200, should be 21672.  Fix? no

Inode 1047956 was part of the orphaned inode list.  IGNORED.
Pass 2: Checking directory structure
Entry 'tmp' in /tmp/systemd-private-bb09aae54cab4e12844e5844d11ca5eb-certbot.service-VSBnVY (685456) has deleted/unused inode 685457.  Clear? no

Entry '1159_key-certbot.pem' in /etc/letsencrypt/keys (930317) has deleted/unused inode 920168.  Clear? no

Entry '1159_key-certbot.pem' in /etc/letsencrypt/keys (930317) has an incorrect filetype (was 1, should be 0).
Fix? no

Entry '1110_csr-certbot.pem' in /etc/letsencrypt/csr (930318) has deleted/unused inode 920176.  Clear? no

Entry '1110_csr-certbot.pem' in /etc/letsencrypt/csr (930318) has an incorrect filetype (was 1, should be 0).
Fix? no

Entry '1106_key-certbot.pem' in /etc/letsencrypt/keys (930317) has deleted/unused inode 920166.  Clear? no

Entry '1106_key-certbot.pem' in /etc/letsencrypt/keys (930317) has an incorrect filetype (was 1, should be 0).
Fix? no

Entry '1109_key-certbot.pem' in /etc/letsencrypt/keys (930317) has deleted/unused inode 920173.  Clear? no

Entry '1109_key-certbot.pem' in /etc/letsencrypt/keys (930317) has an incorrect filetype (was 1, should be 0).
Fix? no

Entry '1146_csr-certbot.pem' in /etc/letsencrypt/csr (930318) has deleted/unused inode 920172.  Clear? no

Entry '1146_csr-certbot.pem' in /etc/letsencrypt/csr (930318) has an incorrect filetype (was 1, should be 0).
Fix? no
...
Pass 3: Checking directory connectivity
Pass 4: Checking reference counts
Inode 685456 ref count is 3, should be 2.  Fix? no

Pass 5: Checking group summary information
Block bitmap differences:  -34565 -(53721--53734) -(59721--59761) -(59981--59983) -(61106--61184) -(61540--61544) -(70964--71007) -(71274--71313) -(84938--84989) -(85084--85107) -(85592--85599) -(116400--116408) -(116423--116436) -(128700--128703) -(128708--128721) -(138904--138914) -(165045--165150) -(169691--169713) -(169717--169742) -(464896--471464) -(471552--471989) -(472928--472947) -(499200--499612) -(501408--501434) -(503808--504070) -(513024--513301) -(513408--513491) -(589477--589480) -(711431--711441) -(747968--748030) -(838733--838740) -(838755--838758) -(838772--838783) -(838791--838800) -(838805--838816) -(838824--838835) -(848384--848972) -(875840--875880) -(1032187--1033031) -(1083840--1083878) -(1120110--1120132) -(1322783--1322975) -(1631196--1631251) -(1635150--1635169) -(1635360--1635391) -(1635571--1635575) -(1635848--1635855) -(1635996--1636001) -1648860 -1648880 -(1715533--1715536) -(1740800--1741311) -(1746432--1746573) -(1750528--1750729) -(1867776--1867880) -(1870717--1871294) -(1880576--1880791) -(1888256--1888258) -1888260 -(1888272--1888273) -(1888275--1888767) -(2226402--2226405) -(2235495--2235719) -(2266304--2266332) -(2301560--2301629) -(2528723--2528753) -(2589088--2589117) -(2597312--2597374) -(2597696--2597757) -(2614784--2615295) -(2619392--2619458) -(2619904--2620297) -2636181 -(2671360--2671491) -(2687328--2687350) -(3068928--3069003) -(3196998--3197002) -(3228728--3228738) -(3236697--3236703) -(3252961--3252970) -(3264276--3264277) -(3264287--3264298) -(3285164--3285170) -(3299518--3299524) -(3399680--3400062) -(3441024--3441129) -(3574080--3574142) -(3601664--3601795) -(3659648--3659724) -(3660672--3660755) -(3704233--3704234) -(3704237--3704242) -3707626 -3708898 -3709310 -3709356 -3709398 -3709984 -(3751694--3751696) -(3751707--3751711) -(3751767--3751768) -(3751774--3751775) -(3751800--3751814) -(3771264--3771343) -(3830025--3830040) -(3860480--3867203) -(3867616--3867644) -(3868160--3868618) -(3869696--3870139) -(4045457--4045483) -(4087936--4088023) -(4088032--4088055) -(4088320--4088780) -(4088960--4089064) -(4089088--4089126) -(4091136--4091324) -(4091392--4092119) -(4092928--4094514) -(4094976--4095854) -(4097088--4097120) -(4097536--4097816) -(4109312--4110157) -(4250368--4250378) -(4278497--4278513) -(4296960--4297014) -(4325486--4325616) -(4325632--4325707) -(4326688--4327074) -(4328826--4328961) -(4329202--4329314) -(4329600--4329666) -(4329764--4329804) -(4332027--4332178) -(4332406--4332476) -(4333568--4333942) -(4334372--4334454) -(4334564--4335227) -(4621153--4621176) -(4669781--4670170) -(4696470--4696548) -(4697074--4697429) -(4697662--4697711) -(4726778--4727894) -(5055921--5056185) -(5056648--5056667) -(5106412--5106620) -(5106668--5107034)
Fix? no

Free blocks count wrong for group #76 (3374, counted=3375).
Fix? no

Free blocks count wrong (445848, counted=445849).
Fix? no

Inode bitmap differences:  -101 -117 -292 -460 -465 -471 -487 -529 -562 -564 -707 -723 -918 -(1837--1838) -2041 -2714 -3593 -3654 -3659 -3894 -3976 -4336 -4425 -5193 -5244 -5252 -5930 -5951 -5967 -(7066--7069) -7431 -8492 -8651 -9298 -9583 -9592 -14261 -14270 -18093 -19214 -21301 -(27843--27844) -27847 -27849 -(27853--27856) -(27868--27869) -(27872--27873) -27875 -27879 -27883 -27885 -(27889--27890) -27892 -162842 -391708 -391741 -391759 -391763 -(391800--391802) -(391804--391805) -(391812--391814) -(391831--391833) -391870 -391873 -391878 -391900 -391902 -(391910--391911) -391915 -391919 -391927 -391956 -392493 -392719 -393759 -393795 -395132 -395134 -395161 -395165 -395221 -395234 -395267 -395289 -(395312--395313) -395315 -395325 -395336 -395387 -395630 -396550 -396589 -(396699--396700) -402594 -(402596--402598) -402601 -(402604--402606) -402608 -(402611--402614) -407918 -413872 -413874 -413881 -413885 -413897 -413900 -413908 -421042 -421202 -421226 -426391 -652905 -(652931--652935) -663035 -685457 -920162 -(920164--920176) -1047956
Fix? no

Directories count wrong for group #84 (17, counted=16).
Fix? no

Free inodes count wrong for group #96 (80, counted=82).
Fix? no

Free inodes count wrong for group #112 (486, counted=487).
Fix? no

Free inodes count wrong (1001361, counted=1001364).
Fix? no


/: ********** WARNING: Filesystem still has errors **********

/: 279759/1281120 files (0.7% non-contiguous), 4673880/5119728 blocks
Nikita Kipriyanov avatar
za flag
This is very strange. Overflow of a file system is not a reason for it to become read-only in Linux. More, it is definitely never a reason for *underlying software array* to become read-only. And it is not 100% full. There must be some other problem. Perform deeper investigation: `dmesg`, `mdadm --detail`, `smartctl` and so on, and share results here. // Also, it is unusual for Linux system to use 20G at root mountpoint during normal operation. Usually those are log files who consume all the space. I operate some systems which are allocated as low as 4G and still having over 2G free in them.
Kamuffel avatar
lu flag
@NikitaKipriyanov I've added some debug information
Nikita Kipriyanov avatar
za flag
Strange to see v0.90 superblocks on array, created in 2017. Also, did you ever checked this FS? Consult with `tune2fs -l /dev/md3`. Perform a read-only check now `e2fsck -n /dev/md3`, and setup regular check (each 10 mounts, maybe).
Kamuffel avatar
lu flag
@NikitaKipriyanov Alright, executed both commands. I've updated my question. Btw, never checked this FS. Had to work on this server (with limited knowledge) setup by a co-worker that left.
Kamuffel avatar
lu flag
@NikitaKipriyanov Any ideas on how to proceed on this matter. My knowledge on FS is extremely limited :).
Score:3
za flag

It is filesystem corruption that caused this switch to read only mode, rather that its overflow, exactly following the mount option errors=remount-ro.

Backup important data and configuration and download them somewhere. Prepare a recovery plan for the case if something important to boot was broken. If it's possible, move important services away to another machine. There'll be some downtime.

I noticed this system doesn't reboot very often (only 7 mounts since 2017, last reboot was in 2019). So I suggest setting up maximal mount count to 1, so it'll be checked each boot:

tune2fs -c 1 /dev/md3

Then reboot. Init script should check and fix the filesystem during boot. However, the corruptions might be so severe so it may require manual interaction, so assure someone is near the server and ready to help you. And, if something important was touched by this corruption, you may have strange problems.

In the worst case, you'll have to reinstall the system. But don't forget setting maximal mount count to 1 again.

Why was the file system corrupted? It just happens. Blocks are stored in memory and the corruption might have taken place there, due to say cosmic rays. Very rare occasion, it happens sometimes. Then, disks are also not ideal and cant detect all errors; there is nonzero bit error rates (look up actual value in your device datasheet), so there is very low chance for data to be read broken, but it is still possible. If this happens on metadata block, the problem may accumulate (file system driver directed by wrong information might do some incorrect assumption and broke the file system further), this is why it is important to check it from time to time.

Kamuffel avatar
lu flag
Thank you for your answer! I will try these steps and look further into it.
mangohost

Post an answer

Most people don’t grasp that asking a lot of questions unlocks learning and improves interpersonal bonding. In Alison’s studies, for example, though people could accurately recall how many questions had been asked in their conversations, they didn’t intuit the link between questions and liking. Across four studies, in which participants were engaged in conversations themselves or read transcripts of others’ conversations, people tended not to realize that question asking would influence—or had influenced—the level of amity between the conversationalists.