feat(integrations): add support for cluster clients from redis sdk #2394

md384 · 2023-09-27T17:36:15Z

This change adds support for cluster clients from the redis sdk (as opposed to the rediscluster library).

This has also been tested in my own app which uses clusters (but not asyncio clusters).

md384 · 2023-10-15T22:10:39Z

I have updated this PR with the db data on the span changes. For redis clusters, I have used the default_node but it might be better to not provide that data at all.

From redis docs - https://redis.readthedocs.io/en/stable/clustering.html#creating-clusters

RedisCluster instance can be directly used to execute Redis commands. When a command is being executed through the cluster instance, the target node(s) will be internally determined. When using a key-based command, the target node will be the node that holds the key’s slot. Cluster management commands and other commands that are not key-based have a parameter called ‘target_nodes’ where you can specify which nodes to execute the command on. In the absence of target_nodes, the command will be executed on the default cluster node. As part of cluster instance initialization, the cluster’s default node is randomly selected from the cluster’s primaries, and will be updated upon reinitialization. Using r.get_default_node(), you can get the cluster’s default node, or you can change it using the ‘set_default_node’ method.

This likely means that it is difficult to get the actual node information and the default node will not always be the one used.

wgiddens · 2023-11-21T20:41:32Z

fixes #2523

szokeasaurusrex

Thank you for your contribution!

I took an initial pass at your PR, and I added a few comments and a question. Please address these, and once I hear back, I will review again

sentry_sdk/integrations/redis/__init__.py

sentry_sdk/integrations/redis/asyncio.py

tests/integrations/redis/asyncio/test_redis_asyncio.py

szokeasaurusrex · 2023-11-22T21:15:23Z

@md384, to get your PR merged as quickly as possible, we would appreciate if you could enable edits from repository maintainers if you are okay with doing so. This way, we can help with any minor changes that need to be made before merging your PR (such as resolving merge conflicts). Thanks!

md384 · 2023-11-28T17:59:19Z

@md384, to get your PR merged as quickly as possible, we would appreciate if you could enable edits from repository maintainers if you are okay with doing so. This way, we can help with any minor changes that need to be made before merging your PR (such as resolving merge conflicts). Thanks!

@szokeasaurusrex I have made your suggested changes and enabled access for maintainers. Thanks!

szokeasaurusrex · 2023-11-28T18:06:05Z

Thank you! I will take a look at your PR again when I have time

szokeasaurusrex

Hey, thank you for addressing my previous suggestions. I have added a few more comments now; once you address them, I think this PR will be ready to merge!

sentry_sdk/integrations/redis/__init__.py

szokeasaurusrex

Thank you for addressing the previous comments. I just tried to resolve the merge conflict between your branch and master, so your PR would be ready to merge, but I was unsure how to resolve it (please see my inline comment).

sentry_sdk/integrations/redis/__init__.py

szokeasaurusrex · 2023-12-04T08:41:59Z

Hi @md384, now that I have resolved the merge conflict between your branch and master, I was able to run our CI checks on your code. It looks like your code is failing our linter ("CI / Lint Sources") and our Redis tests. The aws_lambda test is also failing for now, but this is expected, since I need to allow it to run separately; once you have fixed all the other tests, I will run the aws_lambda test.

Please fix the "CI / Lint Sources" and Redis tests, and let me know if you need help with anything!

…nitialization

md384 · 2023-12-04T21:09:48Z

tests/integrations/redis/cluster/test_redis_cluster.py

+    # on initializing a RedisCluster, a COMMAND call is made - this is not important for the test
+    # but must be accounted for
+    assert len(crumbs) <= 2
+    if len(crumbs) == 2:
+        assert event["breadcrumbs"]["values"][0]["message"] == "COMMAND"


As I mention in the comment, initializing the RedisCluster executes a COMMAND call to the redis server to get available commands. This only happens on the first test run of a parameterized run (more of a problem for test_rediscluster_basic), perhaps there is some caching mechanism somewhere

I would rewrite the assertions as follows:

Suggested change

# on initializing a RedisCluster, a COMMAND call is made - this is not important for the test

# but must be accounted for

assert len(crumbs) <= 2

if len(crumbs) == 2:

assert event["breadcrumbs"]["values"][0]["message"] == "COMMAND"

# on initializing a RedisCluster, a COMMAND call is made - this is not important for the test

# but must be accounted for

assert len(crumbs) in (1, 2)

assert len(crumbs) == 1 || crumbs[0]["message"] == "COMMAND"

The first assertion should be changed because the test should explicitly verify that at least one breadcrumb was created; the previous assertion would have succeeded with no breadcrumbs and the test would only fail later with an IndexError.

Regarding the second assertion, I think my suggested revision conveys more clearly the intent to verify that either there is only one breadcrumb, or if there are two, then the first breadcrumb should be a command.

If you prefer the structure with the if statement for the second assertion, it is okay with me to keep that structure, but in either case, please replace event["breadcrumbs"]["values"][0]["message"] with crumbs[0]["message"]

szokeasaurusrex

I have a few comments on your most recent changes, but I think this is getting really close

szokeasaurusrex · 2023-12-05T12:20:22Z

sentry_sdk/integrations/redis/__init__.py

+        span,
+        # the AsyncClusterPipeline has always had a `_client` attr but it is private so potentially problematic and mypy
+        # does not recognize it - see https://github.com/redis/redis-py/blame/v5.0.0/redis/asyncio/cluster.py#L1386
+        async_redis_cluster_pipeline_instance._client,  # type: ignore[attr-defined]


Is there a way we could try to avoid using _client here, and instead access what we need through a public API? I would strongly prefer to avoid using an external library's private API, since the private API could change at any time and cause problems in the future.

No, I don't believe there is a way we can avoid using _client. We can use a try/except here to be safe and log when it is unavailable. Is there a standard way to raise sentry internal warnings?

Okay, in that case, we can use capture_internal_exceptions here. I also checked that one of your unit tests is verifying this function, so we will get a test failure if _client ever gets removed

szokeasaurusrex · 2023-12-05T12:30:32Z

tests/integrations/redis/cluster/test_redis_cluster.py

+    # on initializing a RedisCluster, a COMMAND call is made - this is not important for the test
+    # but must be accounted for
+    assert len(crumbs) <= 2
+    if len(crumbs) == 2:
+        assert event["breadcrumbs"]["values"][0]["message"] == "COMMAND"


I would rewrite the assertions as follows:

Suggested change

# on initializing a RedisCluster, a COMMAND call is made - this is not important for the test

# but must be accounted for

assert len(crumbs) <= 2

if len(crumbs) == 2:

assert event["breadcrumbs"]["values"][0]["message"] == "COMMAND"

# on initializing a RedisCluster, a COMMAND call is made - this is not important for the test

# but must be accounted for

assert len(crumbs) in (1, 2)

assert len(crumbs) == 1 || crumbs[0]["message"] == "COMMAND"

The first assertion should be changed because the test should explicitly verify that at least one breadcrumb was created; the previous assertion would have succeeded with no breadcrumbs and the test would only fail later with an IndexError.

Regarding the second assertion, I think my suggested revision conveys more clearly the intent to verify that either there is only one breadcrumb, or if there are two, then the first breadcrumb should be a command.

If you prefer the structure with the if statement for the second assertion, it is okay with me to keep that structure, but in either case, please replace event["breadcrumbs"]["values"][0]["message"] with crumbs[0]["message"]

szokeasaurusrex · 2023-12-05T12:34:42Z

tests/integrations/redis/cluster/test_redis_cluster.py

+    if len(crumbs) == 2:
+        assert event["breadcrumbs"]["values"][0]["message"] == "COMMAND"
+
+    crumb = event["breadcrumbs"]["values"][-1]


Suggested change

crumb = event["breadcrumbs"]["values"][-1]

crumb = crumbs[-1]

szokeasaurusrex · 2023-12-05T12:35:29Z

tests/integrations/redis/cluster/test_redis_cluster.py

+    # on initializing a RedisCluster, a COMMAND call is made - this is not important for the test
+    # but must be accounted for
+    assert len(spans) <= 2
+    if len(spans) == 2:
+        assert event["spans"][0]["description"] == "COMMAND"


My previous comment applies here, as well

sentry_sdk/integrations/redis/__init__.py

Co-authored-by: Matthieu Devlin <matt@zumper.com>

szokeasaurusrex

Looks good! Thanks for the contribution!

szokeasaurusrex · 2023-12-07T14:19:08Z

~~It looks like some of the tests are failing; can you please attempt to fix them? Let me know if you have trouble debugging them, and I can help~~

Never mind, the failing test is known to be flaky, and we have a PR to fix it.

md384 force-pushed the m.redis-cluster branch from 5463798 to f798d60 Compare October 3, 2023 15:09

md384 force-pushed the m.redis-cluster branch from f798d60 to ef97545 Compare October 15, 2023 22:04

feat(integrations): add support for cluster clients from redis sdk

ae5bca6

md384 force-pushed the m.redis-cluster branch from ef97545 to ae5bca6 Compare October 16, 2023 17:24

wgiddens mentioned this pull request Nov 21, 2023

Support official redis cluster client #2523

Closed

sentrivana linked an issue Nov 22, 2023 that may be closed by this pull request

Support official redis cluster client #2523

Closed

szokeasaurusrex reviewed Nov 22, 2023

View reviewed changes

fix: review round 1

8734b01

szokeasaurusrex self-assigned this Nov 28, 2023

szokeasaurusrex reviewed Nov 29, 2023

View reviewed changes

sentry_sdk/integrations/redis/__init__.py Outdated Show resolved Hide resolved

sentry_sdk/integrations/redis/__init__.py Show resolved Hide resolved

md384 added 2 commits November 29, 2023 10:36

fix: explicit is not None checks

1f125c1

fix: explicit is not None checks, take 2

06ab2c2

md384 requested a review from szokeasaurusrex November 29, 2023 18:38

szokeasaurusrex reviewed Nov 30, 2023

View reviewed changes

sentry_sdk/integrations/redis/__init__.py Show resolved Hide resolved

md384 and others added 2 commits November 30, 2023 10:27

fix: add try/except to _set_db_data

cd2441a

Merge branch 'master' into m.redis-cluster

5c99ddd

md384 added 2 commits December 4, 2023 12:20

fix: handle additional spans and breadcrumbs caused by rediscluster i…

3e6e9b6

…nitialization

fix: typing for redis integration

c868892

md384 commented Dec 4, 2023

View reviewed changes

md384 requested a review from szokeasaurusrex December 4, 2023 21:11

szokeasaurusrex reviewed Dec 5, 2023

View reviewed changes

fix: simplify assertions

78c8795

md384 commented Dec 6, 2023

View reviewed changes

sentry_sdk/integrations/redis/__init__.py Outdated Show resolved Hide resolved

add capture_internal_exceptions

d5c5a99

Co-authored-by: Matthieu Devlin <matt@zumper.com>

Merge branch 'master' into m.redis-cluster

3360168

szokeasaurusrex approved these changes Dec 7, 2023

View reviewed changes

szokeasaurusrex added the Trigger: tests using secrets PR code is safe; run CI label Dec 7, 2023

rerun CI

7a77138

szokeasaurusrex enabled auto-merge (squash) December 7, 2023 14:14

github-actions bot removed the Trigger: tests using secrets PR code is safe; run CI label Dec 7, 2023

szokeasaurusrex disabled auto-merge December 7, 2023 14:15

szokeasaurusrex added the Trigger: tests using secrets PR code is safe; run CI label Dec 7, 2023

github-actions bot removed the Trigger: tests using secrets PR code is safe; run CI label Dec 7, 2023

Merge branch 'master' into m.redis-cluster

eeb534a

szokeasaurusrex added the Trigger: tests using secrets PR code is safe; run CI label Dec 7, 2023

szokeasaurusrex enabled auto-merge (squash) December 7, 2023 15:33

szokeasaurusrex merged commit 75f89b8 into getsentry:master Dec 7, 2023
470 of 472 checks passed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

feat(integrations): add support for cluster clients from redis sdk #2394

feat(integrations): add support for cluster clients from redis sdk #2394

md384 commented Sep 27, 2023 •

edited by szokeasaurusrex

Loading

md384 commented Oct 15, 2023

wgiddens commented Nov 21, 2023

szokeasaurusrex left a comment •

edited

Loading

szokeasaurusrex commented Nov 22, 2023

md384 commented Nov 28, 2023

szokeasaurusrex commented Nov 28, 2023

szokeasaurusrex left a comment

szokeasaurusrex left a comment

szokeasaurusrex commented Dec 4, 2023

md384 Dec 4, 2023

szokeasaurusrex Dec 5, 2023

szokeasaurusrex left a comment

szokeasaurusrex Dec 5, 2023

md384 Dec 5, 2023

szokeasaurusrex Dec 7, 2023

szokeasaurusrex Dec 5, 2023

szokeasaurusrex Dec 5, 2023

szokeasaurusrex Dec 5, 2023

szokeasaurusrex left a comment

szokeasaurusrex commented Dec 7, 2023 •

edited

Loading

	crumb = event["breadcrumbs"]["values"][-1]
	crumb = crumbs[-1]

feat(integrations): add support for cluster clients from redis sdk #2394

feat(integrations): add support for cluster clients from redis sdk #2394

Conversation

md384 commented Sep 27, 2023 • edited by szokeasaurusrex Loading

md384 commented Oct 15, 2023

wgiddens commented Nov 21, 2023

szokeasaurusrex left a comment • edited Loading

Choose a reason for hiding this comment

szokeasaurusrex commented Nov 22, 2023

md384 commented Nov 28, 2023

szokeasaurusrex commented Nov 28, 2023

szokeasaurusrex left a comment

Choose a reason for hiding this comment

szokeasaurusrex left a comment

Choose a reason for hiding this comment

szokeasaurusrex commented Dec 4, 2023

Choose a reason for hiding this comment

Choose a reason for hiding this comment

szokeasaurusrex left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

szokeasaurusrex left a comment

Choose a reason for hiding this comment

szokeasaurusrex commented Dec 7, 2023 • edited Loading

md384 commented Sep 27, 2023 •

edited by szokeasaurusrex

Loading

szokeasaurusrex left a comment •

edited

Loading

szokeasaurusrex commented Dec 7, 2023 •

edited

Loading