-
Notifications
You must be signed in to change notification settings - Fork 4
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
6.2 upgrade checklist #37
Comments
I marked the optimization to skip over sstables that are fully covered by range tombstones as dropped due to time constraints. I wasn't able to come up with a clean implementation of |
Ack. Ok. |
I was able to compile on Mac without feeaa41. Not sure what to do with it - maybe try dropping it and see if anyone else's build breaks? It is hard to submit upstream right now since I can't say what it fixes. |
Were you able to link the Cockroach binary without that patch, or just
build RocksDB? If I recall correctly things don’t go wrong until you try to
link Cockroach.
…On Thu, Jun 27, 2019 at 2:24 PM Andrew Kryczka ***@***.***> wrote:
I was able to compile on Mac without feeaa41
<feeaa41>.
Not sure what to do with it - maybe try dropping it and see if anyone
else's build breaks? It is hard to submit upstream right now since I can't
say what it fixes.
—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub
<#37?email_source=notifications&email_token=AAGXSICI3L4SXOOSTLTIRVLP4UAVVA5CNFSM4H3WYHH2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODYX66QQ#issuecomment-506457922>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AAGXSIAOPHK73DJRKG5UCY3P4UAVVANCNFSM4H3WYHHQ>
.
|
Thanks Nikhil :). It did appear to successfully link the cockroach binary. I could have made a mistake or could have been using different dependency versions (I was running on qemu on Linux as I don't have a real Mac). Let me try soliciting someone who actually uses Mac to help test. |
Ah, gotcha. So if, using the current master, I run
But perhaps that's been fixed somehow in RocksDB 6.2! |
I see. Thanks for the build output. The candidate 6.2 release branch is here if you want to try: https://github.com/cockroachdb/rocksdb/tree/crl-release-6.2.1 |
I did try! But it looks like some changes are needed in lib roach, too? Is
there a corresponding patch for cockroachdb/cockroach that you've got handy?
…On Thu, Jun 27, 2019 at 2:41 PM Andrew Kryczka ***@***.***> wrote:
I see. Thanks for the build output. The candidate 6.2 release branch is
here if you want to try:
https://github.com/cockroachdb/rocksdb/tree/crl-release-6.2.1
—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub
<#37?email_source=notifications&email_token=AAGXSIGF27QCXYGTOEY72ZDP4UCWNA5CNFSM4H3WYHH2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODYYAPPY#issuecomment-506464191>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AAGXSICKOHL62JHTZEYGUXTP4UCWNANCNFSM4H3WYHHQ>
.
|
Oh really? Do you mind sharing the compiler error? I am building cockroach master branch against that rocksdb 6.2 branch and it seems to work. |
Never mind. TIL |
D'oh. Perhaps the rule for deleting So your patch fixes the libroach compilation failure—thanks! I tested building the latest master with RocksDB pointed at crl-release-6.2.1 and got a jemalloc related error:
That error is indeed resolved by adding feeaa41 to the RocksDB submodule, albeit a bit modified for some upstream changes: diff --git a/db/malloc_stats.cc b/db/malloc_stats.cc
index bcee5c3fb..1dfe0d55b 100644
--- a/db/malloc_stats.cc
+++ b/db/malloc_stats.cc
@@ -20,10 +20,6 @@ namespace rocksdb {
#ifdef ROCKSDB_JEMALLOC
-#ifdef JEMALLOC_NO_RENAME
-#define malloc_stats_print je_malloc_stats_print
-#endif
-
typedef struct {
char* cur;
char* end;
diff --git a/port/jemalloc_helper.h b/port/jemalloc_helper.h
index 0c216face..e5472ff70 100644
--- a/port/jemalloc_helper.h
+++ b/port/jemalloc_helper.h
@@ -9,6 +9,7 @@
#ifdef __FreeBSD__
#include <malloc_np.h>
#else
+#define JEMALLOC_MANGLE
#include <jemalloc/jemalloc.h>
#endif
The rationale, if you're interested in upstreaming this patch, is that the current RocksDB code is broken when linking against a jemalloc that's been configured with a non-empty Anyway, it seems like it's worth working with jemallocs that have been built with |
Understood, thanks a lot for your help! Will submit upstream. |
Heh, thanks for doing all the heavy lifting! |
Release branch: https://github.com/cockroachdb/rocksdb/tree/crl-release-6.2.1 Checklist: cockroachdb/rocksdb#37 Fixes cockroachdb#37602 Release note: None
38576: c-deps: upgrade rocksdb to 6.2.1 r=ajkr a=ajkr Release branch: https://github.com/cockroachdb/rocksdb/tree/crl-release-6.2.1 See checklist: cockroachdb/rocksdb#37 Fixes #37602 Release note: None Co-authored-by: Andrew Kryczka <andrew.kryczka2@gmail.com>
Release branch: https://github.com/cockroachdb/rocksdb/tree/crl-release-6.2.1 Checklist: cockroachdb/rocksdb#37 Fixes cockroachdb#37602 Release note: None
38781: c-deps: upgrade rocksdb to 6.2.1 r=ajkr a=ajkr This is a resubmission of c505411 which got accidentally reverted in f8aaa90. Release branch: https://github.com/cockroachdb/rocksdb/tree/crl-release-6.2.1 Checklist: cockroachdb/rocksdb#37 Fixes #37602 Release note: None Co-authored-by: Andrew Kryczka <andrew.kryczka2@gmail.com>
checklist generation command:
branch: https://github.com/cockroachdb/rocksdb/tree/crl-release-6.2.1
checklist:
--with-jemalloc-prefix
facebook/rocksdb#5521)fallocate
even if hole-punching unsupported Original upstream commit 186b3af is presenttotal_order_seek
for scans Original upstream commit 5a5c049 is presentSyncClosedLogs
for multiple CFs (OnlySyncClosedLogs
for multiple CFs facebook/rocksdb#4460) Original upstream commit a478682 is presentThe text was updated successfully, but these errors were encountered: