I'm having some amount of `PhabricatorApplicationTransactionPublishWorker` tasks leased because workers reached memory limit and are killed (limit is quite high, ~20Gb).
Tried execute tasks via CLI `bin/worker execute --id 123 --trace` with no limit and noticed interesting behavior.
1. Task is executed until forever, memory consumption grows. I've canceled it later.
2. Task is running following queries:
```
...
>>> [28] <query> SELECT `user`.* FROM `user` user WHERE (user.phid IN ('PHID-USER-nsnnowxztidyfw62namu', 'PHID-USER-pgi5dylkzf4n74u5trn5', 'PHID-USER-eh6b6mexpadgyyhhkhfr', 'PHID-USER-6ttizi6pnrvjkaauwb75', ...
<<< [28] <query> 16,978 us
>>> [29] <query> SELECT `user`.* FROM `user` user WHERE (user.phid IN ('PHID-USER-7szdg4elqqaowpzjrkbi', 'PHID-USER-pgi5dylkzf4n74u5trn5', 'PHID-USER-eh6b6mexpadgyyhhkhfr', 'PHID-USER-6ttizi6pnrvjkaauwb75', ...
<<< [29] <query> 17,692 us
>>> [30] <query> SELECT `user`.* FROM `user` user WHERE (user.phid IN ('PHID-USER-nsnnowxztidyfw62namu', 'PHID-USER-pgi5dylkzf4n74u5trn5', 'PHID-USER-eh6b6mexpadgyyhhkhfr', 'PHID-USER-6ttizi6pnrvjkaauwb75', ...
<<< [30] <query> 18,085 us
>>> [31] <query> SELECT `user`.* FROM `user` user WHERE (user.phid IN ('PHID-USER-7szdg4elqqaowpzjrkbi', 'PHID-USER-pgi5dylkzf4n74u5trn5', 'PHID-USER-eh6b6mexpadgyyhhkhfr', 'PHID-USER-6ttizi6pnrvjkaauwb75', ...
<<< [31] <query> 17,795 us
...
```
If you noticed, even and odd queries are repeating. I've canceled task after noticed that because it didn't make sense to perform it any longer.
This might be our data issue but maybe we could debug the reason somehow. Do you have any idea why this is happening?