HBASE-25902 Add missing CFs in meta during HBase 1 to 2 Upgrade #3417
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
🎊 +1 overall
This message was automatically generated. |
|
🎊 +1 overall
This message was automatically generated. |
|
🎊 +1 overall
This message was automatically generated. |
Apache9
reviewed
Jun 24, 2021
Contributor
Apache9
left a comment
Apache9
left a comment
There was a problem hiding this comment.
Is it possible to write a UT for this?
Or have you tried this on real cluster?
| .setColumnFamily(FSTableDescriptors.getTableFamilyDesc(conf)) | ||
| .setColumnFamily(FSTableDescriptors.getReplBarrierFamilyDesc()) | ||
| .build(); | ||
| long pid = this.modifyTable(TableName.META_TABLE_NAME, () -> newMetaDesc, |
Contributor
There was a problem hiding this comment.
This procedure could be done without ClusterSchemaService?
Contributor
Author
There was a problem hiding this comment.
Yes, apparently this can be done without ClusterSchemaService as per my dev testing.
Contributor
Author
|
@Apache9 Unfortunately, writing UT might not be possible (or might require another level of screwup by UT, I tried but looks too complicated to achieve this), but I have tested this change on pseudo-distributed mode (1 ZK server, 1 NN, 1 DN, 1 HM, 4 RS). |
saintstack
reviewed
Jun 28, 2021
| this.tableStateManager.start(); | ||
| } catch (NoSuchColumnFamilyException e) { | ||
| if (tableFamilyDesc == null && replBarrierFamilyDesc == null) { | ||
| LOG.info("For missing CFs in meta, this Exception is expected", e); |
Contributor
There was a problem hiding this comment.
You think operator will know what 'missing CFs in meta' is about? Would it be better to talk about migration from hbase1 to hbase2?
| // initialized) at the expense of bypassing few important tasks as part | ||
| // of active master init routine. So now we abort active master so that | ||
| // next active master init will not face any issues and all mandatory | ||
| // services will be started during master init phase. |
| // of active master init routine. So now we abort active master so that | ||
| // next active master init will not face any issues and all mandatory | ||
| // services will be started during master init phase. | ||
| throw new IOException("Stopping active master after missing CFs are " |
Contributor
There was a problem hiding this comment.
We are not stopping, we are aborting?
Would it be better to throw a PleaseRestartMeException here? (Would have to create it...).
| tries--; | ||
| } | ||
| if (tries <= 0) { | ||
| throw new IOException( |
Contributor
There was a problem hiding this comment.
Rather than new IOE... throw a new HBaseIOE?
virajjasani
force-pushed
the
HBASE-25902-branch-2.4
branch
from
6e871c5 to
fe6323e
Compare
saintstack
approved these changes
Jun 29, 2021
|
🎊 +1 overall
This message was automatically generated. |
anoopsjohn
reviewed
Jun 29, 2021
| } | ||
| } | ||
|
|
||
| public static ColumnFamilyDescriptor getTableFamilyDesc( |
Contributor
There was a problem hiding this comment.
This is a META table CF. Do we have any other util class or so (specific for META) where we can include this?
At least the name should make it clear. This is actually table state CF in Meta.
Contributor
Author
There was a problem hiding this comment.
Let me fix the name. Reg the class to keep it, maybe it is fine here because it is primarily used by FSTD only. Just for upgrade case, master needs it. But I am open to keeping it in other util class if there is better suggestion.
| try { | ||
| this.tableStateManager.start(); | ||
| } catch (NoSuchColumnFamilyException e) { | ||
| if (tableFamilyDesc == null && replBarrierFamilyDesc == null) { |
Contributor
There was a problem hiding this comment.
Seems 2.0.x had extra table state CF only. 2.1.x Had this replication barrier. When table state CF is missing it causes the startup issue right?
When its 2.0.x to 2.3.x upgrade, the repBarrier will get auto created?
Contributor
Author
There was a problem hiding this comment.
This PR is for 2.3+ releases meaning if we come from HBase 1 to 2.3+, the transition should be seamless. Hence, for this upgrade case, both table and repl_barrier will be missing.
When its 2.0.x to 2.3.x upgrade, the repBarrier will get auto created?
I have not tried this one. If table CF is handled, I guess repl_barrier too would have been handled. As per @saintstack's testing, the cluster had to come from 1.2 to earlier than 2.3 release and from that release, it went on to 2.3, hence I am assuming HBase 2 (< 2.3) to 2.3 upgrade should have been smooth. @saintstack thoughts?
Contributor
There was a problem hiding this comment.
I've run various versions of hbase-2.0-hbase2.2.x upgrades successfully... It was when I tried to go from an hbase1.2 version straight to hbase2.3 that I ran into this issue.
Otherwise, agree w/ your thinking around repl_barrier and table CF. Is there a case that one might be in place but not the other -- I don't know-- and does the code do right thing(I've not checked)
Contributor
Author
There was a problem hiding this comment.
Is there a case that one might be in place but not the other -- I don't know-- and does the code do right thing(I've not checked)
Exactly, the case where either 'table' or 'repl_barrier' is missing might be weird case and should not happen given that HBase 2 upgrades are not going to face missing CF issues. That's why I kept && here instead of || to make sure we are specifically handling HBase 1 to 2.3+ upgrade case.
FYI @anoopsjohn
| // Set master as 'initialized'. | ||
| setInitialized(true); | ||
|
|
||
| if (tableFamilyDesc == null && replBarrierFamilyDesc == null) { |
Contributor
There was a problem hiding this comment.
In case one is not there also, need to create?
Contributor
Author
There was a problem hiding this comment.
I was thinking that too but for HBase 1 to 2.3+ upgrade, both will be null for sure so with this case, we are targeting this specific upgrade case.
|
🎊 +1 overall
This message was automatically generated. |
|
🎊 +1 overall
This message was automatically generated. |
|
🎊 +1 overall
This message was automatically generated. |
|
🎊 +1 overall
This message was automatically generated. |
|
🎊 +1 overall
This message was automatically generated. |
virajjasani
added a commit
that referenced
this pull request
Jul 1, 2021
… (#3417) Signed-off-by: Michael Stack <[email protected]>
virajjasani
added a commit
that referenced
this pull request
Jul 1, 2021
Signed-off-by: Michael Stack <[email protected]>
virajjasani
added a commit
that referenced
this pull request
Jul 1, 2021
Signed-off-by: Michael Stack <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
5 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
No description provided.