Skip to content
This repository has been archived by the owner on Apr 26, 2024. It is now read-only.

Add health check endpoint #8048

Merged
merged 6 commits into from
Aug 7, 2020
Merged

Add health check endpoint #8048

merged 6 commits into from
Aug 7, 2020

Conversation

erikjohnston
Copy link
Member

@erikjohnston erikjohnston commented Aug 7, 2020

This gets added under /health on every HTTP listener. Requests to /health get logged at DEBUG rather than INFO.

Question: Where is the best place to document this?

Fixes #8017

@erikjohnston erikjohnston requested a review from a team August 7, 2020 10:07
tests/rest/test_health.py Outdated Show resolved Hide resolved
@clokep
Copy link
Member

clokep commented Aug 7, 2020

This gets added under /health on every HTTP endpoint.

Not 100% sure what this means. I think you mean for each enabled resource?

@clokep
Copy link
Member

clokep commented Aug 7, 2020

Question: Where is the best place to document this?

Probably near the reverse proxy information?

Co-authored-by: Patrick Cloke <clokep@users.noreply.github.com>
@erikjohnston
Copy link
Member Author

This gets added under /health on every HTTP endpoint.

Not 100% sure what this means. I think you mean for each enabled resource?

Woops, yes I mean on every configured HTTP listener

@clokep
Copy link
Member

clokep commented Aug 7, 2020

Requests to /health get logged at INFO rather than DEBUG.

The code here seems to say the opposite:

log_level = logging.INFO if (self.path != b"/health") else logging.DEBUG

I think the code is right and the PR description is wrong (based on the documentation added)?

@erikjohnston
Copy link
Member Author

Requests to /health get logged at INFO rather than DEBUG.

The code here seems to say the opposite:

log_level = logging.INFO if (self.path != b"/health") else logging.DEBUG

I think the code is right and the PR description is wrong (based on the documentation added)?

This is just embarrassing now. Yes, the description is wrong.

docs/reverse_proxy.md Outdated Show resolved Hide resolved
Co-authored-by: Patrick Cloke <clokep@users.noreply.github.com>
@erikjohnston erikjohnston merged commit 7620912 into develop Aug 7, 2020
@erikjohnston erikjohnston deleted the erikj/health_checks branch August 7, 2020 13:21
@michaelkaye
Copy link
Contributor

Wondering if we should add this to a HEALTHCHECK line in the docker files.

I have mixed feelings about re-invoking curl each time the healthcheck fires to ensure the container is happy, but it seems to be the way to do it.

@richvdh
Copy link
Member

richvdh commented Aug 7, 2020

related: #7995

reivilibre added a commit that referenced this pull request Aug 13, 2020
Synapse 1.19.0rc1 (2020-08-13)
==============================

Removal warning
---------------

As outlined in the [previous release](https://github.com/matrix-org/synapse/releases/tag/v1.18.0), we are no longer publishing Docker images with the `-py3` tag suffix. On top of that, we have also removed the `latest-py3` tag. Please see [the announcement in the upgrade notes for 1.18.0](https://github.com/matrix-org/synapse/blob/develop/UPGRADE.rst#upgrading-to-v1180).

Features
--------

- Add option to allow server admins to join rooms which fail complexity checks. Contributed by @lugino-emeritus. ([\#7902](#7902))
- Add an option to purge room or not with delete room admin endpoint (`POST /_synapse/admin/v1/rooms/<room_id>/delete`). Contributed by @dklimpel. ([\#7964](#7964))
- Add rate limiting to users joining rooms. ([\#8008](#8008))
- Add a `/health` endpoint to every configured HTTP listener that can be used as a health check endpoint by load balancers. ([\#8048](#8048))
- Allow login to be blocked based on the values of SAML attributes. ([\#8052](#8052))
- Allow guest access to the `GET /_matrix/client/r0/rooms/{room_id}/members` endpoint, according to MSC2689. Contributed by Awesome Technologies Innovationslabor GmbH. ([\#7314](#7314))

Bugfixes
--------

- Fix a bug introduced in Synapse v1.7.2 which caused inaccurate membership counts in the room directory. ([\#7977](#7977))
- Fix a long standing bug: 'Duplicate key value violates unique constraint "event_relations_id"' when message retention is configured. ([\#7978](#7978))
- Fix "no create event in auth events" when trying to reject invitation after inviter leaves. Bug introduced in Synapse v1.10.0. ([\#7980](#7980))
- Fix various comments and minor discrepencies in server notices code. ([\#7996](#7996))
- Fix a long standing bug where HTTP HEAD requests resulted in a 400 error. ([\#7999](#7999))
- Fix a long-standing bug which caused two copies of some log lines to be written when synctl was used along with a MemoryHandler logger. ([\#8011](#8011), [\#8012](#8012))

Updates to the Docker image
---------------------------

- We no longer publish Docker images with the `-py3` tag suffix, as [announced in the upgrade notes](https://github.com/matrix-org/synapse/blob/develop/UPGRADE.rst#upgrading-to-v1180). ([\#8056](#8056))

Improved Documentation
----------------------

- Document how to set up a client .well-known file and fix several pieces of outdated documentation. ([\#7899](#7899))
- Improve workers docs. ([\#7990](#7990), [\#8000](#8000))
- Fix typo in `docs/workers.md`. ([\#7992](#7992))
- Add documentation for how to undo a room shutdown. ([\#7998](#7998), [\#8010](#8010))

Internal Changes
----------------

- Reduce the amount of whitespace in JSON stored and sent in responses. Contributed by David Vo. ([\#7372](#7372))
- Switch to the JSON implementation from the standard library and bump the minimum version of the canonicaljson library to 1.2.0. ([\#7936](#7936), [\#7979](#7979))
- Convert various parts of the codebase to async/await. ([\#7947](#7947), [\#7948](#7948), [\#7949](#7949), [\#7951](#7951), [\#7963](#7963), [\#7973](#7973), [\#7975](#7975), [\#7976](#7976), [\#7981](#7981), [\#7987](#7987), [\#7989](#7989), [\#8003](#8003), [\#8014](#8014), [\#8016](#8016), [\#8027](#8027), [\#8031](#8031), [\#8032](#8032), [\#8035](#8035), [\#8042](#8042), [\#8044](#8044), [\#8045](#8045), [\#8061](#8061), [\#8062](#8062), [\#8063](#8063), [\#8066](#8066), [\#8069](#8069), [\#8070](#8070))
- Move some database-related log lines from the default logger to the database/transaction loggers. ([\#7952](#7952))
- Add a script to detect source code files using non-unix line terminators. ([\#7965](#7965), [\#7970](#7970))
- Log the SAML session ID during creation. ([\#7971](#7971))
- Implement new experimental push rules for some users. ([\#7997](#7997))
- Remove redundant and unreliable signature check for v1 Identity Service lookup responses. ([\#8001](#8001))
- Improve the performance of the register endpoint. ([\#8009](#8009))
- Reduce less useful output in the newsfragment CI step. Add a link to the changelog section of the contributing guide on error. ([\#8024](#8024))
- Rename storage layer objects to be more sensible. ([\#8033](#8033))
- Change the default log config to reduce disk I/O and storage for new servers. ([\#8040](#8040))
- Add an assertion on `prev_events` in `create_new_client_event`. ([\#8041](#8041))
- Add a comment to `ServerContextFactory` about the use of `SSLv23_METHOD`. ([\#8043](#8043))
- Log `OPTIONS` requests at `DEBUG` rather than `INFO` level to reduce amount logged at `INFO`. ([\#8049](#8049))
- Reduce amount of outbound request logging at `INFO` level. ([\#8050](#8050))
- It is no longer necessary to explicitly define `filters` in the logging configuration. (Continuing to do so is redundant but harmless.) ([\#8051](#8051))
- Add and improve type hints. ([\#8058](#8058), [\#8064](#8064), [\#8060](#8060), [\#8067](#8067))
@clokep
Copy link
Member

clokep commented Aug 24, 2020

Wondering if we should add this to a HEALTHCHECK line in the docker files.

I have mixed feelings about re-invoking curl each time the healthcheck fires to ensure the container is happy, but it seems to be the way to do it.

Looks like #8147 implements this.

babolivier pushed a commit that referenced this pull request Sep 1, 2021
* commit '7620912d8':
  Add health check endpoint (#8048)
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

Provide a http healthcheck endpoint that doesn't log at INFO
4 participants