mailman-2.1.29_6, FreeBSD 12.0-RELEASE-p10
After an unscheduled power cut by our hosting provider, one of the mailing lists hosted on this server came back up with a config.pck file containing 0 bytes. The file's timestamp corresponded to the time of the power loss.
As a result, bin commands such as 'list_lists' would return a corrupt database error, and overnight cron jobs wouldn't run. This was only for one list; other lists on the same server were OK.
I've restored the config.pck file from a backup and things seem to be OK now.
Just checking, though:
Have others encountered config.pck corruption from power loss?
(putting on my tinfoil paranoid hat) Is there any chance the file corruption occurred due to an intrusion attempt, and not the power loss?
Thanks!
dn
On 10/18/19 3:46 PM, David Newman wrote:
I've restored the config.pck file from a backup and things seem to be OK now.
If there was a config.pck.tmp.hhh.ppp with (hhh is the host name and ppp is the PID of the process doing the updating) or a config.pck.last with appropriate size and time stamp, those would have been good and more recent that your backup.
The process for saving a list is
write config.pck.tmp.hhh.ppp this is the updated config.pck remove config.pck.last hard link config.pck.last to config.pck now the previous config.pck has two names rename config.pck.tmp.hhh.ppp to config.pck finally the new file is named config.pck
See the definition of _save() in Mailman/MailList.py
Just checking, though:
- Have others encountered config.pck corruption from power loss?
Yes, there have been reports on this list.
- (putting on my tinfoil paranoid hat) Is there any chance the file corruption occurred due to an intrusion attempt, and not the power loss?
Extremely unlikely. Much more likely is that particular list was in the process of being saved when the power went off.
-- Mark Sapiro <mark@msapiro.net> The highway is for gamblers, San Francisco Bay Area, California better use your sense - B. Dylan
participants (2)
-
David Newman
-
Mark Sapiro