I just encountered this problem on a site I was working on. I tried to batch delete over 40,000 nodes, and in the process the site became unstable and started to run out of memory. The batch delete failed. As a result thousands of "csm_node_temp_XXX" entries were left in the variables table. From that point on the site became terribly unstable, constantly running out of memory. We tracked it down to these thousands of variables, each one containing quite a lot of data: an entire serialized node in each one.

I understand the need to preserve state, since many admins may be working on things at the same time, but with batch delete this state preservation backfires. Perhaps with batch-delete csm could use a special variable that has the form_id of the batch delete form associated with it. I will play around and see if I can find a way to work around the problem.

Comments

thinkyhead’s picture

Priority: Normal » Minor
Status: Active » Closed (cannot reproduce)

Sorry, meant to post this on "csm" not here.