OperationalError: database or disk is full

[jsd@…]

I replied to a trac ticket email, and it didn't work. The command aborted with error 1 so I got a mailer-daemon reply for that. In syslog I see:

Dec 22 10:32:47 ip-10-166-254-128 email2trac thismoment: subject: u'Re: [thismoment] #15002: Make email2trac work'
Dec 22 10:32:47 ip-10-166-254-128 email2trac thismoment: Traceback (most recent call last):
Dec 22 10:32:47 ip-10-166-254-128 email2trac thismoment:   File "/usr/local/bin/email2trac", line 2571, in ?     tktparser.parse(sys.stdin)
Dec 22 10:32:47 ip-10-166-254-128 email2trac thismoment:   File "/usr/local/bin/email2trac", line 1714, in parse     if not self.ticket_update(m, result.group('reply')[:-1], spam_msg):
Dec 22 10:32:47 ip-10-166-254-128 email2trac thismoment:   File "/usr/local/bin/email2trac", line 1010, in ticket_update     tkt.save_changes(self.author, body_text, when, None, str(cnum))
Dec 22 10:32:47 ip-10-166-254-128 email2trac thismoment:   File "build/bdist.linux-i686/egg/trac/ticket/model.py", line 289, in save_changes
Dec 22 10:32:47 ip-10-166-254-128 email2trac thismoment:   File "build/bdist.linux-i686/egg/trac/db/api.py", line 78, in transaction_wrapper
Dec 22 10:32:47 ip-10-166-254-128 email2trac thismoment: OperationalError: database or disk is full

My disk is not full, and adding comments through the web interface works normally, so I assume the database isn't full either.

# df
Filesystem           1K-blocks      Used Available Use% Mounted on
/dev/sda1            103212320  48780892  49189700  50% /
/dev/sdf              51593181   8160798  40810943  17% /var/svnbak
none                    890952         0    890952   0% /dev/shm

[bas]

What trace version do you use? and what is your database backend (sqlite,postgress, ..)? Maybe your database is corrupt. If it is sqlite you can dump and restore it again. I have seen and we also had these kind of errors. Or your log file is too big (trac.log)

[anonymous]

Trac version 0.12.2

Sqlite backend

I don't have a trac.log file apparently (find /srv/trac/thismoment/ -name trac.log returns nothing)

If the sqlite was corrupt, wouldn't normal trac be broken too? The web interface works just fine. I rebuilt the trac db just to be safe (dump/restore) and it's the same. If I run email2trac on the machine then it works. It's just getting the email through postfix it doesn't work.

[bas]

Can you set the python eggcache to a directory that is writeable for everybody, eg:

• /tmp or /var/tmp

Add this option to run_email2trac:

• --eggcache /var/tmp ...

[anonymous]

Setting eggcache had no effect.

[bas]

Ticket is really old. will close it reopen if problem still exists