Since last Ubuntu kernel update (actual 14.04.1 LTS with GNU/Linux 3.13.0-39-generic i686) on my server I'm noticing something strange. My system boots normally - read-writable, no problems. But after a while (day or two) it becomes "Read-only file system", and I can't change files or folders anymore. I have 3 HDD mounted into different directories, and just one of them becomes read-only - the one which is SATA2 connected through an PCIE adapter card. The other two HDD are SCSI and working fine.

dmesg lists some info for sda/sda1 which is the problematic HDD. I'm not sure how to interpret them. The HDD in question I'm using, is SATA2 through PCIE adapter, not ata3.00. It worked before the kernel update. Is there something I can try out?

[335508.395387] ata3.00: exception Emask 0x0 SAct 0x0 SErr 0x0 action 0x0<br />
[335508.395393] ata3.00: BMDMA stat 0x44<br />
[335508.395398] ata3.00: failed command: READ DMA EXT<br />
[335508.395406] ata3.00: cmd 25/00:08:28:13:81/00:00:3e:00:00/e0 tag 0 dma 4096 in<br />
[335508.395406]          res 51/40:08:28:13:81/00:00:3e:00:00/e0 Emask 0x9 (media error)<br />
[335508.395411] ata3.00: status: { DRDY ERR }<br />
[335508.395414] ata3.00: error: { UNC }<br />
[335508.408263] ata3.00: configured for UDMA/100<br />
[335508.408283] sd 2:0:0:0: [sda] Unhandled sense code<br />
[335508.408287] sd 2:0:0:0: [sda]<br />
[335508.408290] Result: hostbyte=DID_OK driverbyte=DRIVER_SENSE<br />
[335508.408293] sd 2:0:0:0: [sda]<br />
[335508.408296] Sense Key : Medium Error [current] [descriptor]<br />
[335508.408301] Descriptor sense data with sense descriptors (in hex):<br />
[335508.408304]         72 03 11 04 00 00 00 0c 00 0a 80 00 00 00 00 00<br />
[335508.408319]         3e 81 13 28<br />
[335508.408326] sd 2:0:0:0: [sda]<br />
[335508.408331] Add. Sense: Unrecovered read error - auto reallocate failed<br />
[335508.408334] sd 2:0:0:0: [sda] CDB:<br />
[335508.408337] Read(10): 28 00 3e 81 13 28 00 00 08 00<br />
[335508.408350] end_request: I/O error, dev sda, sector 1048646440<br />
[335508.408366] ata3: EH complete<br />
[335508.408381] EXT4-fs warning (device sda1): __ext4_read_dirblock:908: error reading directory block (ino 32768161, block 0)
  • 3
    Backup that disk if you haven't already
    – casey
    Nov 16, 2014 at 23:37
  • The HDD was just defect. Thanks casey, I moved some data to save it. After detaching/replacing the HDD everything went good.
    – Hristo
    Dec 20, 2016 at 23:13

2 Answers 2


You should look at the output of dmesg and at the /var/log/syslog to see when this switch is happening and what messages preced it.

I have seen this happen with drives that had intermittent access problems, particularly with USB drives that had bad connectors/cables.

  • I revised the question with some dmesg output. Is there anything I can do/check to find out if some I/O module mess is happening?
    – Hristo
    Nov 16, 2014 at 22:11
  • 1
    @Hristo I would first use a smartctl -a /dev/sda on that drive to see if has recurring errors.
    – Anthon
    Nov 16, 2014 at 22:18
  • smartctl does not detect any errors. I dismounted the drive and used fsck on it. fsck returned that the access time is too short, maybe a zero partition. Hm, it looks like the HDD could be damaged. But on the other hand, it started exactly after I upgraded the kernel. Coincidence?!
    – Hristo
    Nov 17, 2014 at 1:36
  • 1
    Hopefully your drive isn't suffering from a HW problem. Have you checked that the SATA cable is connected properly at each end, and that the adaptor card is seated properly in its slot? My SATA cable occasionally becomes slightly loose at the MB for no clear reason, but a little bit of judicious jiggling soon fixes it. Usually. It's been fine for months but now that I've mentioned it, it'll probably misbehave in the next few days. :)
    – PM 2Ring
    Nov 17, 2014 at 5:58

If you are on a dual boot system and getting the error related to the read-only file system. One probable cause could be that your other system is not correctly shut down and might be in hibernation or sleep. So login to that system and shut it down properly and religion to your Linux system. this resolved the issue in my system

  • 1
    Hello! The question is about an issue on a server machine. It is quite unlikely that a server is a dual boot system because it is intended to be accessed remotely. Nov 21, 2020 at 7:47
  • 1
    Sorry didn't get your intended part, I faced this issue and this was my resolution. Why can't a system be dual boot? I have windows as well installed in my system. Giving a down vote even when this could be a solution is not correct. Nov 22, 2020 at 8:29
  • Did you run into your issue on a server? A typical server runs continuously to provide services to clients. Dual booting it would serve no purpose because those services would not be provided when the system is booted into another OS. That would be an unusual setup, which is not part of the question. Your answer was probably downvoted to show that the proposed solution is not useful, so that others could look on solutions, which likely to help instead. Nov 22, 2020 at 8:52
  • This solution was not based on the client or application server. I ran with this problem on my personal system, and I used this solution to resolve the issue. I posted this because the solution to this issue I was not able to find over net, and I somehow find the error which was coming while mounting the filesystems and then I understand the issue and resolved it. In the question above it was nowhere mention the question is specific to an application or client-server. And I believe I'm not the only person that is using dual boot in the system, and so to help them out I shared my solution Nov 23, 2020 at 0:22
  • You can edit your question to improve it, so it is not necessary to provide explanations in comments. You could just wait till someone will find your solution and upvote it, but I am concerned that people looking for a solution to the question posted above will not find it relevant. For comparison, here is a similar answer, which was upvoted: unix.stackexchange.com/a/420268/423679 Nov 23, 2020 at 7:45

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .