Solaris 10 - Filesystem become read only suddenly

UNIX Solaris
1

Hi,

i just started few days ago to face the problem when the whole system is getting freeze. By freeze i mean:

  • server is reachable (via network)
  • i cannot connect to server via SSH or Telnet(when try to use SSH or telnet it will find server, try to connect, but it freeze after successfull enter the username and password - using root account)
  • i can connect sometimes to the system via the WINSCP i can browse the filesystem but i can`t change any files - when i try it ,the WINSCP get freeze and i have to close it forcely
  • the only way how to get the system working is only to restart it directly by pressing the poweron/off button on the server, then the server boot up and is working normally

Iam not the experienced admin, maybe better to say user who has server and use it mainly for web app and programming, so i do not know where to start to find out where is the problem. What i did is:

  • i checked the harddisk for errors - nothing (using iostat)
  • i checked the /var/adm/messages for some errors but nothing i found (let say, no errors, only notices
  • what seems to me is that the system suddenly become readonly, but why i do not know.

Is there somebody who:

  • maybe experienced the same problem and/or know where could be the problem?
    *can help me to lead what, where and how should i check if there is HW/OS problem?
    *can give me some good advice what to do?

Thank you for your help.

2

The first thing to check is whether the root filesystem is (nearly) full?

Post the output of:

# df -v

If it is nearly full search this forum for "Solaris root full" or similar.

3

Hi , thank you for the reply. The problem is that iam not able to connect to the system so i cannot execute any command. After the restart the root file system is occupied by 49% and after two three hours of checking the occupied size is not growing up

4

This reminds me some issues I had years ago with an AIX system running lotus notes...
The system would at one point be unreacheable via telnet ( not using ssh at the time...) , though could be pinged etc... I used to shutdown HW the box, till decided to always keep a connection on it, after noticing that the Notes admin still had control, asking him to restart Notes solved the issue in most cases and in the other, freed resource so I could connect and reboot gracefully...
The trouble is I dont know if winscp shows the system error messages...
You could be running out of resource ( nproc ) either in terms of number of processes or sockets... you would need to monitor a bit your system to see what is going on, like that we cant say much more...

5

Can you boot to runlevel S? single user.
You have to attach a terminal (A Windows box running a terminal emulator is okay) to the serial port before you boot this way.