- A brief description of what you are looking to do
- How you think this will help
- Why this matters to you
23 results found
profiler - add filter by "appName"
Using the profiler to narrow-down slow operations is great. One thing sorely missing is the ability to filter by "appName" - some of our top slow queries are from dev team working with compass or running experimental code vs the database as opposed to our main production app. We want to be able to ignore these when we perform the analysis - should be simple to implement as the metadata in the profile statistics already has a field with the desired information - "appName"9 votes
Rejecting Index Suggestions
Some Index Suggestions don't work. The Performance Advisor doesn't sample data, and therefore can not predict if some field is worthwhile in an index or not (e.g. some fields may contain the same value for all documents).
If we decide to not apply the Index Suggestion, or we apply it, test it, and remove the Index again because it's not being used by MongoDB, the Index Suggestion is still showing up. It can also trigger alerts.
Please add the possibility to reject an index suggestion, so that it's hidden from the UI and can not trigger alerts.5 votes
Allow Performance Advisor reporting export
Having the ability to extract/export the Performance Advisor report. This would allow users to share a file with others who may not have access to Mongo Atlas.5 votes
Global overview of Index Suggestions
We run more than 60 clusters today on Atlas, and will probably hit 100 clusters in the foreseeable future.
The Index Suggestion alerts are not dependable for us (see https://feedback.mongodb.com/forums/924145-atlas/suggestions/44243568-better-alert-trigger-for-index-suggestions ).
Going through many clusters manually to check them for Index Suggestions is an extremely time-consuming process. We would like to have a global overview of all open Index Suggestions, or at least an overview on a project-basis.4 votes
Determine index's impact score before creating
Currently in Atlas there is no way to test an index's performance before creation. Performance advisor provides an index impact score for suggested indexes. Requesting a feature where you can upload a proposed index and performance advisor can determine the index's impact score. This would help test an index performance before committing to a build4 votes
Recommendation to the change the tier cluster
Based on CPU and memory usage metrics over a certain period of time, a recommendation could be made to change the cluster tier to a higher or lower one (inform which cluster it would be recommended to change to) and notify through some channel, such as eg email/slack.3 votes
View TOP N queries per CPU / memory / disk and number of execution for each query.
This type of metric is common in relational databases and I think it's important to deploy in the Atlas interface.
Top queries by CPU/memory and disk over a period of time or real time. It is important to have this visibility and improve the respective query(s).
It is good to know how many times each query was executed during the respective analyzed period.3 votes
Increase # of Collections in Performance Advisor
Currently the recommendations are for the 20 most active collections. It would be helpful if there was an option to run this for all collections or to set a larger number of collections.3 votes
Use rolling index build by default
We added a few indexes and support mentioned it is not advisable to use the backgroud option when adding an index yet this isthe default setting iwhen adding an index in the performance advisor
Should give download CSV option for getting all the slow queries details
It will be better if a have an option to download all the slow query related information as whatever we see in the Atlas UI.2 votes
increasing the data retention of slow queries to a week.
Currently atlas storing only 24 hours data for slow queries in Profiler. If it enable for a week atleast it will be very helpful for users to track the slow queries in the last week by sorting the query execution time.2 votes
update the course videos the setting up video is completely different from the current atlas environment
update the course videos the setting up video is completely different from the current atlas environment2 votes
Performance Advisor Case Insensitive Queries and $lookup Recommendations - Include Relevant Queries
Currently the Performance Advisor features suggestions for two distinct situations:
* Queries Using regex With Case Insensitivity.
* Queries Using $lookup.
Currently the query or queries that triggered the suggestions are not shared within the Performance Advisor message which requires an extra effort in finding them either in the MongoDB Logs or using the Performance Advisor. As such, it would be useful to attach the relevant query or queries that triggered the suggestions to the Performance Advisor details.2 votes
API for slowQueryLogs should give response in a proper json format like in UI
API for slowQueryLogs should give response in a proper json format like in UI rather than giving raw logs1 vote
Always show shard name on cluster's profiler page
On a cluster’s detail page, there is the Profiler tab. If you click on that, you are brought to the slow query profiler page. What’s not apparent from this view is that this page is not scoped to be cluster wide, but instead is scoped just to the primary mongo node. If you go back to the Overview tab, click on a specific shard on the left and then go to the profiler tab, you see how now, the title is shard specific and truly shows what you’re looking at? Please always display this shard name regardless of the breadcrumb to get there, e.g. if going from the cluster view, display the primary shard name.
I know this title thing seems small, but it caused me to open a support ticket because I didn’t see any of my slow queries show up as they were set to all run on secondary shards. I'm assuming this is probably true for other people.
On a cluster’s detail page, there is the Profiler tab. If you click on that, you are brought to the slow query profiler page. What’s not apparent from this view is that this page is not scoped to be cluster wide, but instead is scoped just to the primary mongo node. If you go back to the Overview tab, click on a specific shard on the left and then go to the profiler tab, you see how now, the title is shard specific and truly shows what you’re looking at? Please always display this shard name regardless of the breadcrumb…1 vote
Show queries that exceeded maxTimeMs in the Atlas Profiler
We had some queries hit the maxTimeMS and fail. We had to download the log files for visibility into those errors. That is info that we would love to see in the Profiler!1 vote
Applying an index suggestion should remove redundant indices automatically
An index suggestion for a compound index should check if it makes an existing index redundant. If so, it should remove it automatically after adding the compound index, or at least add the option to do so.1 vote
profiler - allow for separation of primary vs. secondary queries
We have our production applications using the primary connection. Our support and ad-hoc queries are going against the secondary connection. I would like for these secondary queries to not interfere with or be able to selectively exclude these queries from the profile. Same goes for performance advisor. Performance Advisor will indicate queries are needed do to some support queries when they are not needed or only needed for a short time.1 vote
Provide insight to slow queries on shared instances (like M5)
Currently, features to give insight to slow/optimisable queries are only supported on non-shared instances (i.e. M10+). Would like to see this for shared instances too.
On a side note, this was possible in the cheaper mlab.com setup, which was forced into Atlas.1 vote
- Don't see your idea?