Mercurial > hg
view tests/test-cache-abuse.t @ 30212:260af19891f2
changegroup: increase write buffer size to 128k
By default, Python defers to the operating system for choosing the
default buffer size on opened files. On my Linux machine, the default
is 4k, which is really small for 2016.
This patch bumps the write buffer size when writing
changegroups/bundles to 128k. This matches the 128k read buffer
we already use on revlogs.
It's worth noting that this only impacts when writing to an explicit
file (such as during `hg bundle`). Buffers when writing to bundle
files via the repo vfs or to a temporary file are not impacted.
When producing a none-v2 bundle file of the mozilla-unified repository,
this change caused the number of write() system calls to drop from
952,449 to 29,788. After this change, the most frequent system
calls are fstat(), read(), lseek(), and open(). There were
2,523,672 system calls after this patch (so a net decrease of
~950k is statistically significant).
This change shows no performance change on my system. But I have a
high-end system with a fast SSD. It is quite possible this change
will have a significant impact on network file systems, where
extra network round trips due to excessive I/O system calls could
introduce significant latency.
author | Gregory Szorc <gregory.szorc@gmail.com> |
---|---|
date | Sun, 16 Oct 2016 13:35:23 -0700 |
parents | b0b85d8695cb |
children | 1cc7c96cad75 |
line wrap: on
line source
Enable obsolete markers $ cat >> $HGRCPATH << EOF > [experimental] > evolution=createmarkers > [phases] > publish=False > EOF Build a repo with some cacheable bits: $ hg init a $ cd a $ echo a > a $ hg ci -qAm0 $ hg tag t1 $ hg book -i bk1 $ hg branch -q b2 $ hg ci -Am1 $ hg tag t2 $ echo dumb > dumb $ hg ci -qAmdumb $ hg debugobsolete b1174d11b69e63cb0c5726621a43c859f0858d7f $ hg phase -pr t1 $ hg phase -fsr t2 Make a helper function to check cache damage invariants: - command output shouldn't change - cache should be present after first use - corruption/repair should be silent (no exceptions or warnings) - cache should survive deletion, overwrite, and append - unreadable / unwriteable caches should be ignored - cache should be rebuilt after corruption $ damage() { > CMD=$1 > CACHE=.hg/cache/$2 > CLEAN=$3 > hg $CMD > before > test -f $CACHE || echo "not present" > echo bad > $CACHE > test -z "$CLEAN" || $CLEAN > hg $CMD > after > diff -u before after || echo "*** overwrite corruption" > echo corruption >> $CACHE > test -z "$CLEAN" || $CLEAN > hg $CMD > after > diff -u before after || echo "*** append corruption" > rm $CACHE > mkdir $CACHE > test -z "$CLEAN" || $CLEAN > hg $CMD > after > diff -u before after || echo "*** read-only corruption" > test -d $CACHE || echo "*** directory clobbered" > rmdir $CACHE > test -z "$CLEAN" || $CLEAN > hg $CMD > after > diff -u before after || echo "*** missing corruption" > test -f $CACHE || echo "not rebuilt" > } Beat up tags caches: $ damage "tags --hidden" tags2 $ damage tags tags2-visible $ damage "tag -f t3" hgtagsfnodes1 Beat up hidden cache: $ damage log hidden Beat up branch caches: $ damage branches branch2-base "rm .hg/cache/branch2-[vs]*" $ damage branches branch2-served "rm .hg/cache/branch2-[bv]*" $ damage branches branch2-visible $ damage "log -r branch(.)" rbc-names-v1 $ damage "log -r branch(default)" rbc-names-v1 $ damage "log -r branch(b2)" rbc-revs-v1 We currently can't detect an rbc cache with unknown names: $ damage "log -qr branch(b2)" rbc-names-v1 --- before * (glob) +++ after * (glob) @@ -1,8 +?,0 @@ (glob) -2:5fb7d38b9dc4 -3:60b597ffdafa -4:b1174d11b69e -5:6354685872c0 -6:5ebc725f1bef -7:7b76eec2f273 -8:ef3428d9d644 -9:ba7a936bc03c *** append corruption