[irods/irods#7984] Document rebalance with partial success, commands to query replicas#424
[irods/irods#7984] Document rebalance with partial success, commands to query replicas#424MartinFlores751 wants to merge 1 commit into
Conversation
…to query replicas
MartinFlores751
left a comment
MartinFlores751
left a comment
There was a problem hiding this comment.
Some comments that come to mind.
|
|
||
| To get the number of data objects with replicas which are not marked "good": | ||
| ``` | ||
| SELECT COUNT(DATA_COLL_ID) WHERE DATA_REPL_STATUS != '1' |
There was a problem hiding this comment.
Not sure if DATA_ID is more appropriate here, or if DATA_COLL_ID is what is best...
There was a problem hiding this comment.
I think DATA_ID is correct - that will match which data objects have non-good replicas...
DATA_COLL_ID will be the COLL_ID of the parent of the data objects in question... so I don't think that's right.
| To find all data objects with replicas which are not marked "good": | ||
| ``` | ||
| SELECT COLL_NAME, DATA_NAME WHERE DATA_REPL_STATUS != '1' | ||
| ``` | ||
|
|
||
| To get the number of data objects with replicas which are not marked "good": |
There was a problem hiding this comment.
Consider removing the quotations around good. That is the proper name of the replica status: https://docs.irods.org/5.0.2/system_overview/data_objects/#replica-status
There was a problem hiding this comment.
I think the quotes are okay - or perhaps you're looking to make them backticks which makes it look like a literal or codename?
I think backticks could be more correct... please look around and compare to other places where we talk about replica statuses.
|
|
||
| When a rebalance completes successfully, the timestamp AVU is removed. | ||
|
|
||
| If the rebalance operation encounters data objects that cannot be replicated, due to being stale or locked, a `-1834000 REBALANCE_NOT_COMPLETE` message will be displayed and a non-zero exit code is returned after replicating all remaining good data objects. Afterwards, the unreplicated data objects must be handled manually if replication is still desired. |
There was a problem hiding this comment.
what non-zero error code? can we know and be explicit here?
| To find all data objects with replicas which are not marked "good": | ||
| ``` | ||
| SELECT COLL_NAME, DATA_NAME WHERE DATA_REPL_STATUS != '1' | ||
| ``` | ||
|
|
||
| To get the number of data objects with replicas which are not marked "good": |
There was a problem hiding this comment.
I think the quotes are okay - or perhaps you're looking to make them backticks which makes it look like a literal or codename?
I think backticks could be more correct... please look around and compare to other places where we talk about replica statuses.
|
|
||
| To get the number of data objects with replicas which are not marked "good": | ||
| ``` | ||
| SELECT COUNT(DATA_COLL_ID) WHERE DATA_REPL_STATUS != '1' |
There was a problem hiding this comment.
I think DATA_ID is correct - that will match which data objects have non-good replicas...
DATA_COLL_ID will be the COLL_ID of the parent of the data objects in question... so I don't think that's right.
2 participants
Issue quick link: irods/irods#7984
This PR documents the new rebalance error from PR irods/irods#8930, along with queries to get data objects that could not be replicated.