feat(logging): Add startup logging for shard counts #25378
Conversation
devanbenz
force-pushed
the
feat/476-startup-progress
branch
from
1ce74a8
to
b4b96c7
Compare
devanbenz
changed the title
|
@devanbenz @davidby-influx I have some hard numbers to put behind my claim that scanning all the directories and files and then scanning them again doesn't take much longer than scanning them once. On my 3.5 year old Influx laptop, there are ~3.5 million files in my home directory. Scanning all of them cold after dropping all the disk caches takes about 50 seconds. Scanning a second time with the cache hot only takes 4 seconds. So scanning them twice only adds 8% beyond scanning them one time. |
|
@gwossum this is ready now, I was thinking I could set up a test suite entry for maybe 0 shards and then it hitting an error to show that the uint64 doesn't go below 0 per the checks I have in the interface but idk if I should take the time to wire that up seems like a lot of test boilerplate I need to create for it. I will also squash the commits and have a single descriptive commit later. |
This PR adds a check to see how many shards are remaining vs how many shards are opened as well as the percentage. closes influxdata/feature-requests#476
devanbenz
force-pushed
the
feat/476-startup-progress
branch
from
11f53c2
to
26de392
Compare
This PR adds a check to see how many shards are remaining vs how many shards are opened as well as the percentage. closes influxdata/feature-requests#476
|
@gwossum I've modified my store_test code to include the changes that you established to fix the additional shard problem. The scaffolding is fixed now. |
This PR adds a check to see how many shards are remaining
vs how many shards are opened as well as the percentage.
closes https://github.com/influxdata/feature-requests/issues/476