Conversation
Ignore errors when setting log retention
Update magerun to 9.3.0
Add Google Tag Manager server-side
…jax-suggest-cache-type feat: DEVOPS-621 add search ajax suggest cache type
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 790d4298e5
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| {% if (gtm_enabled | default(false)) %} | ||
| if (req.http.Host == "{{ mageops_gtm_domain }}" ) { | ||
| set req.backend_hint = gtm_director.backend(); | ||
| return (pass); |
There was a problem hiding this comment.
Preserve forwarded client IP before GTM passthrough
The new GTM host branch returns pass immediately, so requests never reach the later req.restarts == 0 block that normalizes X-Forwarded-For from X-Real-IP for localhost-origin traffic. In the standalone nginx→varnish setup, this means GTM can receive loopback/internal addresses instead of the real client IP, which breaks IP-dependent tagging/geo logic specifically when req.http.Host matches the GTM domains.
Useful? React with 👍 / 👎.
…he-configuration feat: DEVOPS-623 add realpath cache configuration
Implement rclone replacement for s3fs
Co-authored-by: Szpadel <[email protected]>
Add slowlog depth trace option to PHP
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: dee72a53d4
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| block: | ||
| - include_tasks: 002-setup-mounts.yml | ||
| - include_tasks: 003-setup-watchdog.yml | ||
| when: s3_mount_backend == 's3fs' |
There was a problem hiding this comment.
Restore watchdog setup for non-s3fs backends
Conditioning 003-setup-watchdog.yml on s3_mount_backend == 's3fs' drops the periodic fix/remount cron jobs for goofys (which previously had them) and for the new rclone backend, so a crashed mount on those backends will now stay broken until manual intervention. This is a functional regression for any environment using s3_mount_backend: goofys after upgrading.
Useful? React with 👍 / 👎.
| case "${BACKEND}" in | ||
| s3fs) | ||
| SOURCE_PATTERN="$(printf '%s' "${SOURCE}" | sed 's/#\+/.*/g')" | ||
| echo "s3fs.*${SOURCE_PATTERN}.*${MOUNTPOINT}" |
There was a problem hiding this comment.
Relax s3fs/goofys kill pattern matching in mount watchdog
For s3fs/goofys, SOURCE_PATTERN already contains the backend token from fstab source (for example s3fs#bucket becomes s3fs.*bucket), but the emitted regex prepends s3fs.*/goofys.* again. That makes the fallback kill path require two backend tokens in the process command line, so stuck mount processes are often not matched and cannot be killed when unmount fails.
Useful? React with 👍 / 👎.
Version check was using unpopulated variables for upgrade detection
Fix plugin reinstall logic
Move lazy resize to etc.php
…search-results feat: DEVOPS-627 allow cache search results
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: fef179bbf1
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| args: | ||
| removes: "/usr/share/opensearch/plugins/{{ item }}" | ||
| loop: "{{ opensearch_plugin_list_command.stdout_lines }}" | ||
| loop: "{{ opensearch_plugins }}" |
There was a problem hiding this comment.
Remove every installed plugin before an OpenSearch upgrade
This loop now only removes entries from opensearch_plugins, so any plugin that was installed on the previous node but has since been dropped from that variable will survive the package upgrade. OpenSearch loads plugins from disk on startup, so a stale plugin from the old version can keep the service from coming back even though the desired plugin set was reduced. The previous implementation enumerated the installed plugins first, which avoided that upgrade failure mode.
Useful? React with 👍 / 👎.
| {% for instance in varnish_backend_instances_app %} | ||
| gtm_director.add_backend(gtm_{{ instance.instance_id | replace('-','') }}); |
There was a problem hiding this comment.
Include extra instances in the GTM directors
The template now defines gtm_* backends for both varnish_backend_instances_app + varnish_backend_instances_extra above, but this director is still populated from varnish_backend_instances_app only (and the preview director below has the same pattern). In environments that rely on the extra ASG for capacity or failover, GTM traffic will never use those healthy extra nodes, and it can go unavailable if the base app pool is empty or unhealthy even though GTM backends were generated for the extra pool.
Useful? React with 👍 / 👎.
No description provided.