can't ALTER TABLE ... FREEZE ... after successfully restore, frozen-metadata.txt present #826
Assignees
Milestone
Comments
This is a trimmed error message, could you share the full error message? |
|
when you store data to backup? Could you check is backuped data still present on remote storage? |
|
@Slach but we have wrapped line in our log viewer. Just these information.
|
In general, backup mechanism should prevent data loss or another chaos problems occur. So we emulate if need to restore data to clickhouse server pod from s3. And once restore completed, we will start to ingest the new data and start backup. And in this step, can not execute backup(create_remote).
@Slach We have checked it, these backup data are still present on remote storage. |
|
if we also click house-server have the below error message Do not sure what is the '\t'... |
|
Could you share results of |
|
@Slach The result for sql command: And our table engine is |
|
@Slach But once expired ttl(previous 1 hour) and stored on s3. When we restore and execute watch command. Original issues problem occur again. Does s3 tier disk not support backup restore after ttl move to s3? |
|
You need just to delete new versions of clickhouse-backup will just ignore this file |
|
@Slach 2.4.28 is released? the latest release version on GitHub repo is 2.4.27 |
|
2.4.27 also contain ignoring fronzen-metadata.txt |
|
@Slach But we originally just use 2.4.27.... Can refer to this issue description |
Slach
changed the title
|
for successfull ALTER TABLE ...FREZZE during create remote please remove |
|
@Slach
|
|
are you sure you deleted all frozen-metadata*.* files? what return? |
@Slach This command returns null response. Should I upgrade to 2.4.29 to double check it whether is solved? |
|
what return what return ALTER TABLE table=ck_job.otel_trace_lab_ubt FREEZE WITH NAME 'test_freeze';
ALTER TABLE table=ck_job.otel_trace_lab_ubt UNFREEZE WITH NAME 'test_freeze';? 2.4.29 is not released yet, only docker image |
|
@Slach
|
|
you didn't delete frozen_metadata.txt |
|
@Slach So I would like to confirm something. When we execute restore command, and clickhouse would freeze table, right? And v2.4.29 will includes this feature? we are worried about these operator too manual so that easily operation error. |
|
2.4.29 will include fixes for create / upload / download / restore commands to allow ignore frozen_metadata, unfortunatelly we can't affect exists unproperly restored data, and it shall be fixed manually |
|
@Slach |
when we create backup then during restore we download it but after it FREEZE command inside 2.4.29 will ignore frozen_metadata during create / upload / download and restore |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Description
After restore data from s3, we want to keep watching to backup data to s3 through backup tool server api.
But we always have this problem..., please follow this below error message.
We are not sure why can not create_remote after restore data and why shows
\tthis error. hope someone can assist to answer it.But if we do not execute restore, just directly execute watch after pod start running. the backup can normally running.
Version
disk configuration of clickhouse server
print-config for clickhouse-backup tool
The text was updated successfully, but these errors were encountered: