fix(bootstrap): wait core tables are ready before copying #183
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
thalesmg
previously approved these changes
Oct 18, 2024
zmstone
reviewed
Oct 19, 2024
| , config = Opts | ||
| }, | ||
| %% Create (or copy) the mnesia table and wait for it: | ||
| ok = create_table(MetaSpec), | ||
| ok = mria_mnesia:copy_table(?schema, Storage), | ||
| %% Ensure replicas are available before starting copy: | ||
| ok = mria_mnesia:wait_for_tables([?schema]), |
There was a problem hiding this comment.
I assume this function does not return anything else than ok?
if that's the case, maybe add a comment here to document what it means to the node boot sequence if this wait has to take a very long time (or never returns ok)
There was a problem hiding this comment.
It can return {error, ...} when mnesia is stopped while the bootstrap is in progress.
This calls mria_mnesia:wait_for_tables/1, which (unlike mnesia:wait_for_tables/1) logs quite extensive diagnostic information every 30 seconds if this wait is taking too long, so this should be enough? A comment detailing why this was needed won't hurt, in addition to commit message, I guess.
In specific circumstances `mria_mnesia:copy_table/2` may fail with
`{system_limit, '$mria_rlog_sync', {Node, none_active}}` error, which
crashes the node.
Consider the following scenario:
1. Node `N1` starts up and bootstraps Mria.
2. Node `N2` starts up and bootstraps Mria.
3. Node `N2` joins cluster consisting of node `N1`.
4. Node `N2` runs `mria_mnesia:join_cluster/1` and starts Mria again.
5. At the exact same time node `N1` decides to restart for some reason.
6. During bootstrap, node `N2` tries to copy `$mria_rlog_sync` table.
7. Mnesia sees there's nowhere to copy from and aborts the operation.
8. Mria fails to start.
While unlikely, in practice this might be achieved when the operator
performs unusual maintenance operations, e.g. simultaneously requests
version upgrade and scales the cluster up.
Silence "expression updates a literal" compiler lint recently introduced in erlang/otp#8069.
keynslug
force-pushed
the
fix/race-system-limit
branch
from
13f9607 to
6084346
Compare
zmstone
approved these changes
Oct 21, 2024
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
In specific circumstances
mria_mnesia:copy_table/2may fail with{system_limit, '$mria_rlog_sync', {Node, none_active}}error, which crashes the node.Consider the following scenario:
N1starts up and bootstraps Mria.N2starts up and bootstraps Mria.N2joins cluster consisting of nodeN1.N2runsmria_mnesia:join_cluster/1and starts Mria again.N1decides to restart for some reason.N2tries to copy$mria_rlog_synctable.While unlikely, in practice this might be achieved when the operator performs unusual maintenance operations, e.g. simultaneously requests version upgrade and scales the cluster up.
Fixes EMQX-13309.