Evergreen History Load Timeout

[StomproLARL2023]

Hello, Commit ce26344 changed how the Evergreen History was loaded to break up the requests sent to Evergreen into batches of 100.

But what I'm observing is that there is still a 60 second timeout for the task of loading all the batches of 100.

An extended timeout doesn't get set until after all the requests to Evergreen are done.

I have a case where a customer has 2455 items in their history. And the request to load that data in Aspen times out after 60 seconds after only getting through 3 batches of 100. First batch takes 27 seconds, next batch takes 29 seconds, and then the 3 batch gets requested right before the script times out at 60 seconds.

Should each batch of 100 also have a set_time_limit(60); call to reset the timeout to 60 seconds for each batch? The request to Evergreen in my case is fast. Each batch of 100 gets processed on the server in .03 seconds when I request locally. So most of the time is in the loop in code/web/Drivers/Evergreen.php in getReadingHistory. I would guess that creating new marc records is what is taking the time.

The first 3 batches of 100 history items for this customer are all missing the source circ data since they are older. So this could also be related to that code that kept those from being skipped also. Maybe it takes significantly more time to deal with those with missing source circ.

We have about 1500 users with more than 500 titles in their circ history, so being able to load that data is important to us.

There was a reference to (Ticket 128963) in the commit message.