Wednesday, July 22, 2009

SHMMAX, SHMALL and the like

This page is really helpful:

http://www.pythian.com/news/245/the-mysterious-world-of-shmmax-and-shmall

Monday, June 29, 2009

The importance of current directory

The commands "rm" and "touch" (there might be others) sometimes need that the current directory is writeable for the current user. I have not seen this to be always true, but in case you are getting a "permission denied" error while executing these commands, and if everything else that's obvious (like write permissions on the target directory / file) looks ok, just try running the command after changing the current directory to one that is writeable for you. /tmp is a simple choice.


For rm failure, the error is more explicit to smell this cause: it says "rm: cannot get current directory, permission denied". For touch failure, I just got "permission denied", and it worked after changing current dir to a writeable one.

Monday, May 25, 2009

Figuring out postfix errors

I recently faced a problem with email from one of our unix nodes. We have a system (a set of bash scripts) which sends out email on completion of a test suite to various mailing lists. One fine night, it just broke - stopped sending email to anyone. On some investigation, on the machine from where emails were to be sent, found this:

mc1 04:52:50 /var/spool/postfix/defer $ pwd
/var/spool/postfix/defer
mc1 04:52:51 /var/spool/postfix/defer $ ls
1 3 4 5 6 7 8 9 A B C D E F
mc1 04:53:29 /var/spool/postfix/defer $ cd 1
mc1 04:53:32 /var/spool/postfix/defer/1 $ ls
14BA8540BB 164E8540C5 191C9540E0 1C77B540C6
mc1 04:53:34 /var/spool/postfix/defer/1 $ cat 14BA8540BB | head

: host relayhost.domain.com[ip appr] said: 452 Insufficient system storage (in reply to RCPT TO command)
recipient=username@domain.com
offset=423
status=4.0.0
action=delayed
reason=host relayhost.domain.com[ip appr] said: 452 Insufficient system storage (in reply to RCPT TO command)

I have of course replaced real machine names and IP addresses with dummies. There were also some "connection refused" messages in the files here, which were probably an offshoot of the root problem.

Haven't found what's to be done yet - will update this when done!