You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
 
 
 

4.0 KiB

layout type title date tags
post handbook Todo List 2014-11-10 todo admin

These are checklist admin needs to get done.

June 10

  • backup PostgreSQL
  • backup and tar MongoDB
  • cleanup everything first (should take < 10 mins for all)
    • backfill_reviews
    • check_student_issue
    • check_teacher_issue
    • archive_issues
    • cleanup_glitches
  • mongo_report --date=20xx0610 --report=yearly
    • It's much better to have 1, 2, 3, ..., 9 separately (much faster)
  • run_misc_support --date=20xx0610
    • it's also a good idea to separate into 1, 2, 3, 4, 5, ..., 9 or it could take > 5 hours to finish
    • with i5 & SSD, 9 takes little over 1:30 hour
  • mongo_yearly_stat --date=20xx0610 all

Pre-compile report for office

on /office/opec/stat/ & /misc_support/ (on ztokyo3 contained enough tools)

  • staticsitegen
    • don't forget to change code for Jun10Renderer & MiscSupportJun10Renderer
    • MiscSupportJun10Renderer alone takes roughly 35 mins on i5 and SSD
  • ./toolkit/__meduza_cleanup.sh
  • ~/bin/sync_psis_meduza.sh
  • double check if data is there,
    • /office/opec/stat/
    • /misc_support/ แบบสรปเบกจร
    • /misc_support/ ภาคเรยนท 1, PDF

At the beginning of the month

If it's March 1st, there is an extra work below.

  • backup PostgreSQL
  • backup and tar MongoDB
  • cleanup everything first (should take < 10 mins for all)
    • backfill_reviews
    • check_student_issue
    • check_teacher_issue
    • archive_issues
    • cleanup_glitches
  • mongo_report --date=20xxyy01 --report=monthly
    • It's much better to have 1, 2, 3, ..., 9 separately (much faster)
  • visit /sub4sum/ page to invoke
    • sub4sum
    • sub4sum_office
  • archive_students_transfer - This usually throws 100k tasks to Celery which takes > 5 hours to complete.

November 10

The process is simple, but it takes time to get jobs done.

  • backup PostgreSQL
  • backup and tar MongoDB
  • run_misc_support --date=20xx1110
    • it's a good idea to separate into 1, 2, 3, 4, 5, ..., 9 or it could take > 5 hours to finish
    • processed on tokyo1 & 3 simultaneously took only < 15 mins to get all things done
    • note load distributes throughout every app & DB servers. No hard hit at all. That's why it runs smoothly fast here.
  • mongo_report --date=20xx1110 --report=ms_nov10
    • 1 takes less than 10 minutes to complete (most work are done in run_misc_support already.
    • note this does take its toll on ztokyo2 though. DB server gets hit hard at 100% every cores. Thus, it's oddly taking longer than run_misc_support in production.

Pre-compile report for office on /misc_support/ too. (on ztokyo3 contained enough tools)

  • staticsitegen -- don't forget to change code for Nov10Renderer
    • This takes > 1:15 to finish on i5 & SSD; meduza is a blocking single process, so it doesn't get any better on ztokyo3, 1:35 it took on Nov 10, 2014
  • ./toolkit/__meduza_cleanup.sh
  • ~/bin/sync_psis_meduza.sh
  • double check if data is there,
    • แบบสรปเบกจรง 10 พย.
    • ภาคเรยนท 2, PDF

February 15

  • backup PostgreSQL
  • backup and tar MongoDB
  • make sure ms_rate is fulfilled
  • mongo_report --date=20xx0215 --report=ms_feb15

Pre-compile report for office on /misc_support/ too. (on ztokyo3 contained enough tools)

  • staticsitegen -- don't forget to change code for Feb15Renderer
  • ./toolkit/__meduza_cleanup.sh
  • ~/bin/sync_psis_meduza.sh
  • double check if data is there,
    • สรปขอเบกประมาณการ

At the end of Febuary

This is an extra work from at the beginning of the month

  • mongo_yearly_stat --date=20xx0215 all
    • It's much better to have 1, 2, 3, ..., 9 separately (much faster)