[GH-ISSUE #213] smtpd gives up after a couple of emails #156

New issue

Closed

opened 2026-02-25 23:41:23 +03:00 by kerem · 4 comments

kerem commented

2026-02-25 23:41:23 +03:00

Owner

Originally created by @danielfaulknor on GitHub (Jan 22, 2019).
Original GitHub issue: https://github.com/healthchecks/healthchecks/issues/213

I'm running smtpd in a docker container:

set -e -x

setcap 'cap_net_bind_service=+ep' /usr/bin/python3.6

su -c "python3 /src/manage.py smtpd --port=25" hc

It connects up fine, and takes a few email pings, then these start hitting the logs. My CONN_MAX_AGE is 0.

I guess it's holding the connection open, and mysqld goes away. This results in emails being rejected.

Originally created by @danielfaulknor on GitHub (Jan 22, 2019). Original GitHub issue: https://github.com/healthchecks/healthchecks/issues/213 I'm running smtpd in a docker container: ```#!/bin/sh set -e -x setcap 'cap_net_bind_service=+ep' /usr/bin/python3.6 su -c "python3 /src/manage.py smtpd --port=25" hc ``` It connects up fine, and takes a few email pings, then these start hitting the logs. My `CONN_MAX_AGE` is 0. ``` error: uncaptured python exception, closing channel <smtpd.SMTPChannel connected 50.31.156.116:35051 at 0x7feda0e9f4e0> (<class 'django.db.utils.OperationalError'>:(2006, 'MySQL server has gone away') [/usr/lib/python3.6/asyncore.py|read|83] [/usr/lib/python3.6/asyncore.py|handle_read_event|423] [/usr/lib/python3.6/asynchat.py|handle_read|171] [/usr/lib/python3.6/smtpd.py|found_terminator|386] [/src/hc/api/management/commands/smtpd.py|process_message|25] [/usr/local/lib/python3.6/dist-packages/django/db/models/manager.py|manager_method|82] [/usr/local/lib/python3.6/dist-packages/django/db/models/query.py|get|393] [/usr/local/lib/python3.6/dist-packages/django/db/models/query.py|__len__|250] [/usr/local/lib/python3.6/dist-packages/django/db/models/query.py|_fetch_all|1183] [/usr/local/lib/python3.6/dist-packages/django/db/models/query.py|__iter__|54] [/usr/local/lib/python3.6/dist-packages/django/db/models/sql/compiler.py|execute_sql|1061] [/usr/local/lib/python3.6/dist-packages/django/db/backends/utils.py|execute|68] [/usr/local/lib/python3.6/dist-packages/django/db/backends/utils.py|_execute_with_wrappers|77] [/usr/local/lib/python3.6/dist-packages/django/db/backends/utils.py|_execute|85] [/usr/local/lib/python3.6/dist-packages/django/db/utils.py|__exit__|89] [/usr/local/lib/python3.6/dist-packages/django/db/backends/utils.py|_execute|85] [/usr/local/lib/python3.6/dist-packages/django/db/backends/mysql/base.py|execute|71] [/usr/lib/python3/dist-packages/MySQLdb/cursors.py|execute|250] [/usr/lib/python3/dist-packages/MySQLdb/connections.py|defaulterrorhandler|50] [/usr/lib/python3/dist-packages/MySQLdb/cursors.py|execute|247] [/usr/lib/python3/dist-packages/MySQLdb/cursors.py|_query|411] [/usr/lib/python3/dist-packages/MySQLdb/cursors.py|_do_query|374] [/usr/lib/python3/dist-packages/MySQLdb/connections.py|query|292])``` I guess it's holding the connection open, and mysqld goes away. This results in emails being rejected.

kerem closed this issue

2026-02-25 23:41:23 +03:00

kerem commented

2026-02-25 23:41:23 +03:00

Author

Owner

@danielfaulknor commented on GitHub (Jan 23, 2019):

If anyone finds this... There's a Lambda function in https://github.com/healthchecks/healthchecks/issues/139

I've switched to using that.

@danielfaulknor commented on GitHub (Jan 23, 2019): If anyone finds this... There's a Lambda function in https://github.com/healthchecks/healthchecks/issues/139 I've switched to using that.

kerem commented

2026-02-25 23:41:23 +03:00

Author

Owner

@cuu508 commented on GitHub (Feb 21, 2019):

I'm wondering if "MySQL server has gone away" happens because of a connection being idle for too long.

Did you also see similar errors when running manage.py sendalerts? sendalerts is similar in that it is also a management command and it also doesn't do anything special with connection handling. (but, it runs queries every few seconds)

@cuu508 commented on GitHub (Feb 21, 2019): I'm wondering if "MySQL server has gone away" happens because of a connection being idle for too long. Did you also see similar errors when running `manage.py sendalerts`? sendalerts is similar in that it is also a management command and it also doesn't do anything special with connection handling. (but, it runs queries every few seconds)

kerem commented

2026-02-25 23:41:23 +03:00

Author

Owner

@danielfaulknor commented on GitHub (Feb 26, 2019):

sendalerts is fine yeah

@danielfaulknor commented on GitHub (Feb 26, 2019): sendalerts is fine yeah

kerem commented

2026-02-25 23:41:23 +03:00

Author

Owner

@cuu508 commented on GitHub (Feb 27, 2019):

So my theory is that the manage.py smtpd is opening a single database connection and expecting to use it indefinitely. If the emails are infrequent then the connection is idle for long periods and gets closed – either by kernel or by mysql, not sure.

I added a workaround to get a new db connection on every email. I'm closing this issue for now, and we'll see if the problem resurfaces.

@cuu508 commented on GitHub (Feb 27, 2019): So my theory is that the `manage.py smtpd` is opening a single database connection and expecting to use it indefinitely. If the emails are infrequent then the connection is idle for long periods and gets closed – either by kernel or by mysql, not sure. I added a workaround to get a new db connection on every email. I'm closing this issue for now, and we'll see if the problem resurfaces.