You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I have automated tests every time I save any file (with mocha --watch), and this means knex store is initialized A LOT throughout my day.
I've noticed that after a few writes, any attempted knex connection to postgres results in this:
Unhandled rejection error: sorry, too many clients already
at Connection.parseE (/app/node_modules/pg/lib/connection.js:606:11)
at Connection.parseMessage (/app/node_modules/pg/lib/connection.js:403:19)
at Socket.<anonymous> (/app/node_modules/pg/lib/connection.js:123:22)
at Socket.emit (events.js:210:5)
at addChunk (_stream_readable.js:308:12)
at readableAddChunk (_stream_readable.js:289:11)
at Socket.Readable.push (_stream_readable.js:223:10)
at TCP.onStreamRead (internal/stream_base_commons.js:182:23)
The cause seems to be, at least partially, that all the session expiry queries persist after the application is shut down. From psql:
somedbname=# select pid, query from pg_stat_activity;
pid | query
------+----------------------------------------------------------------------------
29 |
1389 | delete from "sessions" where expired < CAST($1 as timestamp with time zone)
1410 | delete from "sessions" where expired < CAST($1 as timestamp with time zone)
1388 | delete from "sessions" where expired < CAST($1 as timestamp with time zone)
624 | select pid, query from pg_stat_activity;
1337 | delete from "sessions" where expired < CAST($1 as timestamp with time zone)
1412 | delete from "sessions" where expired < CAST($1 as timestamp with time zone)
1419 |
1270 | delete from "sessions" where expired < CAST($1 as timestamp with time zone)
1338 | delete from "sessions" where expired < CAST($1 as timestamp with time zone)
1404 | delete from "sessions" where expired < CAST($1 as timestamp with time zone)
1405 | delete from "sessions" where expired < CAST($1 as timestamp with time zone)
1390 | delete from "sessions" where expired < CAST($1 as timestamp with time zone)
1429 | delete from "sessions" where expired < CAST($1 as timestamp with time zone)
1367 | delete from "sessions" where expired < CAST($1 as timestamp with time zone)
1408 |
1430 | delete from "sessions" where expired < CAST($1 as timestamp with time zone)
1234 | delete from "sessions" where expired < CAST($1 as timestamp with time zone)
1368 | delete from "sessions" where expired < CAST($1 as timestamp with time zone)
1413 | delete from "sessions" where expired < CAST($1 as timestamp with time zone)
1241 | delete from "sessions" where expired < CAST($1 as timestamp with time zone)
1414 | delete from "sessions" where expired < CAST($1 as timestamp with time zone)
1415 | delete from "sessions" where expired < CAST($1 as timestamp with time zone)
1416 | delete from "sessions" where expired < CAST($1 as timestamp with time zone)
1335 | delete from "sessions" where expired < CAST($1 as timestamp with time zone)
1417 |
1418 |
1421 |
1264 | delete from "sessions" where expired < CAST($1 as timestamp with time zone)
1277 | delete from "sessions" where expired < CAST($1 as timestamp with time zone)
647 | delete from "sessions" where expired < CAST($1 as timestamp with time zone)
1420 |
651 | delete from "sessions" where expired < CAST($1 as timestamp with time zone)
1248 | delete from "sessions" where expired < CAST($1 as timestamp with time zone)
1422 |
1393 | delete from "sessions" where expired < CAST($1 as timestamp with time zone)
1301 | delete from "sessions" where expired < CAST($1 as timestamp with time zone)
1423 | delete from "sessions" where expired < CAST($1 as timestamp with time zone)
1395 | delete from "sessions" where expired < CAST($1 as timestamp with time zone)
1396 | delete from "sessions" where expired < CAST($1 as timestamp with time zone)
1397 | delete from "sessions" where expired < CAST($1 as timestamp with time zone)
1321 | delete from "sessions" where expired < CAST($1 as timestamp with time zone)
1424 | delete from "sessions" where expired < CAST($1 as timestamp with time zone)
1425 |
1398 | delete from "sessions" where expired < CAST($1 as timestamp with time zone)
1426 | delete from "sessions" where expired < CAST($1 as timestamp with time zone)
1351 | delete from "sessions" where expired < CAST($1 as timestamp with time zone)
1382 | delete from "sessions" where expired < CAST($1 as timestamp with time zone)
1399 | delete from "sessions" where expired < CAST($1 as timestamp with time zone)
1427 |
1327 | delete from "sessions" where expired < CAST($1 as timestamp with time zone)
1401 | delete from "sessions" where expired < CAST($1 as timestamp with time zone)
1403 | delete from "sessions" where expired < CAST($1 as timestamp with time zone)
1407 |
1354 | delete from "sessions" where expired < CAST($1 as timestamp with time zone)
1431 | delete from "sessions" where expired < CAST($1 as timestamp with time zone)
Is there a workaround for this?
I don't know if this should be fixed on connect-session-knex side, or I should be doing something differently from my code.
While this gets sorted, does anyone know how to remove those queries programmatically, ideally efficiently? My best guess was select pid, query from pg_stat_activity where query like '%sessions%expired%' and pid != pg_backend_pid() and pg_cancel_backend(pid), but it seems pg_cancel_backend is ran before the pid != pg_backend_pid check, so the command never finishes and a lot of those cookie clearing queries persist.
My understanding of postgres is quite limited.
The text was updated successfully, but these errors were encountered:
This query is slightly better. Doesn't make the connection tests error out, but it does spam a lot of Connection Error: Connection ended unexpectedly in the mocha console
select pid, query from pg_stat_activity where pg_stat_activity.state like 'idle' and query like '%session%expired%' and pg_terminate_backend(pid)
I have automated tests every time I save any file (with
mocha --watch
), and this means knex store is initialized A LOT throughout my day.I've noticed that after a few writes, any attempted knex connection to postgres results in this:
The cause seems to be, at least partially, that all the session expiry queries persist after the application is shut down. From psql:
Is there a workaround for this?
I don't know if this should be fixed on connect-session-knex side, or I should be doing something differently from my code.
While this gets sorted, does anyone know how to remove those queries programmatically, ideally efficiently? My best guess was
select pid, query from pg_stat_activity where query like '%sessions%expired%' and pid != pg_backend_pid() and pg_cancel_backend(pid)
, but it seemspg_cancel_backend
is ran before thepid != pg_backend_pid
check, so the command never finishes and a lot of those cookie clearing queries persist.My understanding of postgres is quite limited.
The text was updated successfully, but these errors were encountered: