somandal opened a new pull request, #16664:
URL: https://github.com/apache/pinot/pull/16664
Today the `segments/{tableName}/metadata` API sends out one request per
segment. For large tables this can be problematic because the
MultiHttpRequest's multi-get function creates a separate thread for each
request. For tables with a very large number of segments we have seen issues
where a huge number of threads are spawned at once and even results in issues
like OOMs
This PR addresses the above by modifying the call to get all segments to
send one request per server
- Adds a server level API to return a list of `SegmentMetadata` objects, one
for each segment. This naturally filters out OFFLINE segments as a segment data
manager won't exist for OFFLINE segments (so no filtering needed on the caller
side)
- Changes the controller API for `segments/{tableName}/metadata` to use the
above server API instead
Verifications done using `HybridQuickStart` with 4 servers:
With these changes, we see only 4 URLs when sending out the request (logs
added locally for testing):
```
2025/08/21 16:07:03.116 ERROR [MultiHttpRequest] [grizzly-http-server-27]
************ NUMBER OF urlsAndRequestBodies: 4
2025/08/21 16:07:03.116 ERROR [MultiHttpRequest] [async-task-thread-60]
*********** submitting request, url:
http://192.168.86.234:7501/tables/airlineStats_OFFLINE/segments/metadata?columns=,
body:
2025/08/21 16:07:03.116 ERROR [MultiHttpRequest] [async-task-thread-26]
*********** submitting request, url:
http://192.168.86.234:7500/tables/airlineStats_OFFLINE/segments/metadata?columns=,
body:
2025/08/21 16:07:03.117 ERROR [MultiHttpRequest] [async-task-thread-28]
*********** submitting request, url:
http://192.168.86.234:7502/tables/airlineStats_OFFLINE/segments/metadata?columns=,
body:
2025/08/21 16:07:03.117 ERROR [MultiHttpRequest] [async-task-thread-58]
*********** submitting request, url:
http://192.168.86.234:7503/tables/airlineStats_OFFLINE/segments/metadata?columns=,
body:
```
Verified that we get back data for all segments (we expect a total of 62
segments as RF = 2, and there are 31 unique segments):
```
2025/08/21 16:06:27.219 ERROR [ServerSegmentMetadataReader]
[grizzly-http-server-26] ***** SIZE of serviceResponse: 62 ********
...
...
2025/08/21 16:07:03.128 ERROR [ServerSegmentMetadataReader]
[grizzly-http-server-27] ***** SIZE of serviceResponse: 4 ********
2025/08/21 16:07:03.128 ERROR [ServerSegmentMetadataReader]
[grizzly-http-server-27] ***** for server:
http://192.168.86.234:7503/tables/airlineStats_OFFLINE/segments/metadata?columns=__1,
number of segment metadata returned: 16 ********
2025/08/21 16:07:03.129 ERROR [ServerSegmentMetadataReader]
[grizzly-http-server-27] ***** for server:
http://192.168.86.234:7500/tables/airlineStats_OFFLINE/segments/metadata?columns=__1,
number of segment metadata returned: 16 ********
2025/08/21 16:07:03.129 ERROR [ServerSegmentMetadataReader]
[grizzly-http-server-27] ***** for server:
http://192.168.86.234:7502/tables/airlineStats_OFFLINE/segments/metadata?columns=__1,
number of segment metadata returned: 15 ********
2025/08/21 16:07:03.130 ERROR [ServerSegmentMetadataReader]
[grizzly-http-server-27] ***** for server:
http://192.168.86.234:7501/tables/airlineStats_OFFLINE/segments/metadata?columns=__1,
number of segment metadata returned: 15 ********
```
The older way sends out 62 separate requests (capturing only a subset for
brevity):
```
2025/08/21 16:06:27.219 ERROR [ServerSegmentMetadataReader]
[grizzly-http-server-26] ***** for server:
http://192.168.86.234:7501/tables/airlineStats_OFFLINE/segments/airlineStats_OFFLINE_16086_16086_0/metadata?columns=__1,
number of segment metadata returned: 1 ********
2025/08/21 16:06:27.219 ERROR [ServerSegmentMetadataReader]
[grizzly-http-server-26] ***** for server:
http://192.168.86.234:7502/tables/airlineStats_OFFLINE/segments/airlineStats_OFFLINE_16076_16076_0/metadata?columns=__1,
number of segment metadata returned: 1 ********
2025/08/21 16:06:27.219 ERROR [ServerSegmentMetadataReader]
[grizzly-http-server-26] ***** for server:
http://192.168.86.234:7503/tables/airlineStats_OFFLINE/segments/airlineStats_OFFLINE_16071_16071_0/metadata?columns=__1,
number of segment metadata returned: 1 ********
2025/08/21 16:06:27.219 ERROR [ServerSegmentMetadataReader]
[grizzly-http-server-26] ***** for server:
http://192.168.86.234:7502/tables/airlineStats_OFFLINE/segments/airlineStats_OFFLINE_16100_16100_0/metadata?columns=__1,
number of segment metadata returned: 1 ********
```
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
