Abhilash Raj a écrit le 09/04/2020 à 20:21 :
On Thu, Apr 9, 2020, at 1:13 AM, Gilles Filippini wrote:
Hello,
I run a small mailman3 server (6 lists, less than 100 users), and I can't access anymore to the administration web page for one of these lists:
Something went wrong Mailman REST API not available. Please start Mailman core.
This happens only on a single list? It is weird.
Can you check Mailman Core's logs?
I checked logs with no clue. The only logs I could use were the postgres'. I saw a ROLLBACK statement after a series of hold emails checks. There weren't many, so I carefully dropped them from the '_request' table, and I could access the web administration pages for this list again afterwards.
The other lists seems OK.
Looking at the postregsql logs I see a full load of these requests (4 to 5 per second!): LOG: statement: SELECT "django_q_ormq"."id", "django_q_ormq"."key", "django_q_ormq"."payload", "django_q_ormq"."lock" FROM "django_q_ormq" WHERE ("django_q_ormq"."key" = 'default' AND "django_q_ormq"."lock" < '2020-04-09T07:57:37.828200+00:00'::timestamptz) LIMIT 1
This is un-related. This is meant for only routine jobs, which are stored in the database. You can switch to using redis or something for this if you want to reduce load on your database.
All the available brokers are listed here1.
I eventually found out this is a known issue [2]. [2] https://lists.mailman3.org/archives/list/mailman-users@mailman3.org/thread/Z...
Because our lists are very low traffic I changed the poll delay to 30s via setting.py.
I'm not at ease playing with other brokers because our mailman3 instance is a dockerized one.
Thanks for your help,
_g.