Adjust timing of aspace2alma processes so we don't miss the Analytics report

regineheberlein commented 3 weeks ago

What maintenance needs to be done?

https://github.com/pulibrary/aspace_helpers/issues/538 suggests that on some days, Analytics runs before the latest data snapshot becomes available to it.

Push the Analytics report (sc_active_barcodes) back by one hour and re-schedule the dependent aspace2alma processes accordingly.

Level of urgency

[x] High
[ ] Moderate
[ ] Low

Why is this maintenance needed?

So the Analytics report doesn't report out old data.

Acceptance criteria

[ ] sc_active_barcodes.csv is current as of 8pm the previous day

Implementation notes, if any

mzelesky commented 3 weeks ago

Just let me know when you're ready for the schedule of the Analytics report to be updated. I would plan on shifting the schedule to 3 AM instead of 1 AM.

regineheberlein commented 3 weeks ago

Thanks @Mark C. @.***>. I’ll let you know by the end of the week.

Regine Heberlein (she/her) Library IT Data Analyst 609-258-6156 @.**@.>

My working day may not be your working day. Please do not feel obliged to reply to this email outside of your regular working hours.

From: mzelesky @.> Date: Wednesday, April 24, 2024 at 1:02 PM To: pulibrary/aspace_helpers @.> Cc: Regine I. Heberlein @.>, Author @.> Subject: Re: [pulibrary/aspace_helpers] Adjust timing of aspace2alma processes so we don't miss the Analytics report (Issue #539)

Just let me know when you're ready for the schedule of the Analytics report to be updated. I would plan on shifting the schedule to 3 AM instead of 1 AM.

— Reply to this email directly, view it on GitHubhttps://github.com/pulibrary/aspace_helpers/issues/539#issuecomment-2075429905, or unsubscribehttps://github.com/notifications/unsubscribe-auth/ABNKKU3DASUSHDJNF6I6VADY67QTTAVCNFSM6AAAAABGTHK7E2VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDANZVGQZDSOJQGU. You are receiving this because you authored the thread.Message ID: @.***>

regineheberlein commented 2 weeks ago

Server Time Considerations

the Alma front-end runs on Eastern time and falls back/springs forward with Eastern Standard (winter time) / Daylight Savings (summer time)
lib-jobs runs on UTC, i.e. Eastern summer time +4 / Eastern winter time + 5
lib-sftp runs on UTC, i.e. Eastern summer time +4 / Eastern winter time + 5
aspace2alma is run from a whenever job on lib-jobs in UTC
Alma captures the state of the data at (circa) 8pm; it becomes available to Analytics at (circa) 1am (not at midnight as we had determined a year ago--just speculating: this may or may not be due to the Alma backend also running on UTC?)

Other Considerations

Alma job schedules are not absolute; jobs can be delayed
When aspace2alma has run concurrently with pulfalight, we've had failures in the past. pulfalight runs on the full hour between 9am-9pm and runs a full index at 10pm, so avoid those times

Current schedule

Alma Analytics is scheduled to run at 1am Eastern
aspace2alma is scheduled to run at 8:30am UTC / lib-jobs time
Alma runs the import job at 9am Eastern

service	server	server clock	scheduled literally as	EDT (summer) (=UTC-4)	EST (winter) (=UTC-5)	UTC (summer)(=EDT+4)	UTC (winter) (=EST+5)	duration (appr.)
Alma Analytics	Alma	Eastern	1:00am	1:00am	1:00am	5:00am	6:00am	30 mins
aspace2alma	lib-jobs	UTC	8:30am (`30 8`)	4:30am	3:30am	8:30am	8:30am	2 hrs 30 mins
Alma Import Job	Alma	Eastern	9:00am	9:00am	9:00am	1:00pm	2:00pm	5 mins

regineheberlein commented 2 weeks ago

If we push the Analytics report back by 2 hrs, i.e. to run at 3am Eastern, it will finish at 7:30 UTC (summer) / 8:30 UTC (winter). aspace2alma runs at 8:30 UTC, so we have a conflict in winter.

So we need to push aspace2alma back correspondingly. It would run at 10:30 UTC and finish at 1pm UTC (summer) / 2pm UTC (winter). The Alma import job runs at 1pm UTC (summer) / 2pm UTC (winter), so we have a conflict year-round.

So we also need to push the Alma import job back correspondingly, which would mean it runs after the start of business, mid-morning at 11am Eastern.

mzelesky commented 2 weeks ago

Do you have concerns about the import profile job running during the workday? It doesn't seem like a problem to me, but I might not understand staff considerations.

regineheberlein commented 2 weeks ago

@mzelesky We are ready for you to re-schedule the sc_active_barcodes job as well as the ASpace to Alma with Items please. Both need to be pushed back by 2 hours, so could you please schedule

sc_active_barcodes job : at 3am
ASpace to Alma with Items : at 11am

Thank you!

mzelesky commented 2 weeks ago

Timings were adjusted as of 9:10 AM on 4/29.

regineheberlein commented 2 weeks ago

The corresponding aspace2alma schedule change is now also in place. Thanks again!

pulibrary / aspace_helpers