DA stopped, System file Read-only, how to backup ?

[koudou]

I had apparently a big problem on a server (centos) and now, the main partition / goes in read-only mode. I passed the hardware test and discovered nothing.

ssh starts, but not DA nor httpd nor exim.

When I start on rescue mode, I can see the directories and files (including /home) and everything seems to be there, but in read-only mode.

I have decided to move everything on a new server, but I cannot backup old server:
- on old server, in rescue mode, I can start directadmin, but queue is not working and backups are not done (see question 1)
- on old server, in normal mode, I cannot start nothing as disk is forbidden in writing (touch, etc...) question 2

QUESTION 1 : Do you know how to start queue of backups or command to issue to simulate a request of backup from DA (even user level is ok : it is more work , but if it can be done, it is only work)

QUESTION 2 : is there a way to force a new install over the existing one without erasing user data ? (do remember that I can write on the disk only if in rescue mode, that is to say with a simplified system)

QUESTION 3 : how to "manually" get all the necessary info (especially mails pop and imap and databases) to be sure to have it working ?

Thank you in advance for your help if you can: this is an operational server and you all know side effects on a dead operational server ...

Michel

 

[tillo]

It seems like that partition's filesystem has a critical error; you could just fix it permanently using a live linux distribution and fsck, that way you won't have to bother with a crippled system and do the migration easily.

 

[koudou]

Thank you tillo for your suggestion.

I also had the idea, but I cannot get a physical access to the server (only distance control) as it is part of a very big farm on racks. I have no CD, no usb connection, and the only choice I have is to restart on a rescue system with my main disk as a disk (and hopefully, this is possible) of put a new system on the disk by deleting all previous data.

Sometimes, life is not so easy ...

 

[tillo]

I know the feeling, I also have a few machines in distant datacenters and I struggle doing something that would take just a minute if I had physical access...

Does this "rescue system" allows you to unmount the root partition? You could fsck the corrupted filesystem from there.
If you can't, you could try just remounting it read-only (or just run the system in normal mode) then run fsck on it anyway, you may be lucky and it won't need it to be unmounted.