SIP2 Patron Information times out on too many checkouts/holds

[Thomas Berezansky]

When a patron has too many items checked out, on hold, or both patron information responses can time out counting them.

I believe this is due to the fact that the backend retrieves full information for each circulation and hold, specifically Titles loaded from parsing MARC, when only the counts are actually being used.

The two branches below provide support to Evergreen and SIPServer for only fetching IDs of the circs/holds, which are then counted instead of the titles. By not loading the titles the patron information response is sped up considerably.

Evergreen Side, to add the ability to fetch just IDs:
http://git.evergreen-ils.org/?p=working/Evergreen.git;a=shortlog;h=refs/heads/user/tsbere/sip2_ids_only

SIPServer Side, to have it ask for just IDs:
http://git.evergreen-ils.org/?p=working/SIPServer.git;a=shortlog;h=refs/heads/user/tsbere/ids_only

[Ben Shum (bshum)]

Adding initial target, but this may be good to backport as it could be bad for older systems too?

[Galen Charlton (gmc)]

I've committed this to SIPServer master and Evergreen master, rel_2_6, and rel_2_5. I've also added a follow-up to ensure that the renew-all message doesn't break if the user neglects to upgrade SIPServer at the same time they upgrade Evergreen.

Thanks, Thomas!

[Galen Charlton (gmc)]

Following up discussion today in IRC regarding the non-typical addition of the $ids_only parameter to ->charged_items() to the middle rather than the end of the parameter list, the patch at the tip of the user/gmcharlt/lp1296937_split_public_and_private_interfaces branch in the working/Evergreen repository splits the public and implementation interfaces for the charged items lookup. As the commit message indicates, this is meant to be tested with a reversion of SIPServer commit c97d64412.

http://git.evergreen-ils.org/?p=working/Evergreen.git;a=shortlog;h=refs/heads/user/gmcharlt/lp1296937_split_public_and_private_interfaces

Consider this a pullrequest for this follow-up.

[Mike Rylander (mrylander)]

This looks like the best way forward. All the benefits of the speedups without depending on a quirk of implementation or reordering of positional params. Thanks for working this out, Galen, and thanks for pushing for a fix, Thomas.

Here's a signed-off branch:

http://git.evergreen-ils.org/?p=working/Evergreen.git;a=shortlog;h=refs/heads/user/miker/lp1296937_split_public_and_private_interfaces

[Jeff Godin (jgodin)]

I like this approach, and think it deals with all of the issues raised and any that I could think of.

I've cherry picked Evergreen commit 80ee141 to master, rel_2_5, and rel_2_6.

My intent was to push a revert of SIPServer commit c97d64412 to master, but I appear to lack the access to push to SIPServer. I'll work with someone to resolve that, then I think we're done with this bug.

Thanks Thomas, Ben, Galen, and Mike!

[Jeff Godin (jgodin)]

Fixed local issue on my end, pushed revert of SIPServer commit c97d64412 to master.

Thanks again, all!