Over the last 2 weeks I’ve had problems on three sites with the exact same problem.

The watchdog table grew FASTER than the cron job would have deleted it.

(the cron job should delete old log entries older than e.g. 3 days)

PLUS the PUBLIC site display was full of error messages, which could not be logged to the database anymore

This again ended in a “device full” last night when one watchdog table became 5 gigs.

(and I had a “table full” problem on another site last week – tough 64bit MYSQL should be able to do 4TB spaces, that one was 4GB only)

Now, it’s obvious that disk full problems shouldn’t be a problem for a content management system like Drupal to take care about,

but the system or application administrators. However, this particular situation IS an application problen inherent to Drupal.

Because, the REASON the table grew so fast was, that the certain table “cache” was marked as crashed/corrupt in mysql,

and hence EVERY httpp request to the Drupal site caused several tries to read/write from/to the cache table – resulting in an error being logged.

One must agree that it makes NO SENSE to dump a “table …. marked as crashed” several hundred times PER SECOND into the watchdog,

and I therefore already urgently requested a redesign of some kind of alert threshold.

So my idea is, if an error message was sent 100 times to the watchdog in the last hour (which could have been seen by a human), chances are the next 120,000 entries in that hour won’t add any information to it, but just fill it up with crap. So I would say we supress this exact same watchdog entry… (yaddaa yadda – it’s loss of useless information – because the table IS still corrupt… no matter if it happens at 1:31 or 1:35)

Another nice feature I think will be required would have various levels definable by a) severity of the error b) component (i.e. PHP vs. applciation errors) where the error arrives.

Whatever solution deems appropriate to the Drupal core maintainers, the current situation of Drupal filling it’s own watchdog with useless entries

leading to a halting server (because mysql tables after often placed on /var partitions, which also carry system log file)

is not an operable solution but a hazard for production systems.

Fortunately the watchdog has been the target of several improvements recently to move Drupal more to an enterprise-class application, so I’ve got bit hopes that this flaw will

find it’s way into Drupal some day as well.

Let me know what you think.

Did your Content Management System ever fill up the whole disk of the server?