storage: remove replica pointer tag #15212
Conversation
|
This change is |
| @@ -513,8 +513,6 @@ func newReplica(rangeID roachpb.RangeID, store *Store) *Replica { | |||
| r.rangeStr.store(0, &roachpb.RangeDescriptor{RangeID: rangeID}) | |||
| // Add replica log tag - the value is rangeStr.String(). | |||
There was a problem hiding this comment.
so comments, such useful. amaze.
|
@tamird Those tests failures are really scary: We didn't upgrade RocksDB recently, did we? So what changed? |
petermattis
force-pushed
the
pmattis/replica-pointer-tag
branch
from
27193ee to
c8d6697
Compare
|
@jordanlewis just reported this failure as well. We did not, and I'm not
sure. I won't have time to investigate until this afternoon due to the
all-hands. Perhaps @benesch has a minute to spend on this.
…On Thu, Apr 20, 2017 at 12:32 PM, Peter Mattis ***@***.***> wrote:
@tamird <https://github.com/tamird> Those tests failures are really scary:
cli.test: /go/src/github.com/cockroachdb/cockroach/c-deps/rocksdb.src/db/inlineskiplist.h:424: int rocksdb::InlineSkipList<Comparator>::RandomHeight() [with Comparator = const rocksdb::MemTableRep::KeyComparator&]: Assertion `height <= kMaxHeight_' failed.
We didn't upgrade RocksDB recently, did we? So what changed?
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#15212 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/ABdsPLfkOIFSMgv84QKkVDPQ6U3cn5Miks5rx4iLgaJpZM4NDRXt>
.
|
|
Current suspicion is that #15188 is the culprit. Attempting to reproduce. |
|
Updated theory: this is probably build skew. We don't bust c-deps caches when we change |
|
Is there a particular motivation for removing this? It's also been useful when looking into replica GC bugs. |
|
Nuked the affected agent; all seems well. I'll work on a way to force rebuilding of C dependencies when |
I thought those bugs were well in the past. Am I mistaken about that? |
|
And the motivation was a minor amount of log cleanliness. |
|
LGTM. I noticed this and was thinking about removing it too. |
They are. I was just pointing out that it's been useful for more than just the original issue. It's not a huge deal to get rid of it again though, we can re-add it as needed. |
I could also comment it out. Preference? |
|
Commenting it out sgtm |
The replica pointer tag was used for debugging cockroachdb#11591, which has been long since fixed.
petermattis
force-pushed
the
pmattis/replica-pointer-tag
branch
from
c8d6697 to
a5bbfcf
Compare
Done. |
The replica pointer tag was used for debugging #11591, which has been
long since fixed.