[jira] [Comment Edited] (IMPALA-12066) If the gathering thread skips table because it cannot acquire a table lock, it may lead to issues with lost some updates.

["Zhi Tang (Jira)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/IMPALA-12066?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17712798#comment-17712798 ] 

Zhi Tang edited comment on IMPALA-12066 at 4/16/23 3:20 PM:
------------------------------------------------------------

The smaller the topic_update_tbl_max_wait_time_ms value, the greater the probability of this occurrence.


was (Author: tangzhi):
The greater the topic_update_tbl_max_wait_time_ms value, the greater the probability of this occurrence.

> If the gathering thread skips table because it cannot acquire a table lock, it may lead to issues with lost some updates.
> -------------------------------------------------------------------------------------------------------------------------
>
>                 Key: IMPALA-12066
>                 URL: https://issues.apache.org/jira/browse/IMPALA-12066
>             Project: IMPALA
>          Issue Type: Bug
>          Components: Catalog
>    Affects Versions: Impala 4.2.0
>            Reporter: Zhi Tang
>            Priority: Major
>
> After IMPALA-6671, if the topic update thread cannot acquire a table lock it skips the table from that topic update. In the CatalogServiceCatalog#lockHdfsTblWithTimeout() method, we think it's safe to skip the table when the gathering thread cannot get a read lock in attempt 2. This is not entirely correct because it's not guarantee that the tbl version will be updated outside current window of the topic updates when the mutating thread fails for some reason.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

---------------------------------------------------------------------
To unsubscribe, e-mail: issues-all-unsubscribe@impala.apache.org
For additional commands, e-mail: issues-all-help@impala.apache.org