Discussion:
【wait on line】weird question about svnsync
Dummy
2017-06-26 11:07:56 UTC
Permalink
dear subversion:
I have a weird question about svnsync:
i svnsync gzrepos(centos 6.4-svn1.6.11) to gz-mirror1(centos 7-svn1.9.5),
when it was done successfully, i found that the mirror repo(gz-mirror1) is more less than source repo(gzrepos) in sizes ,the source repo(gzrepos) is 177G while the mirror one(gz-mirror1) is only 66G
for example: revision:2012,it seems to be compressed

svnsync logs:



however, the mirror repo(gz-mirror1) seems OK, we can access,checkout,show log etc.
i heard that subversion 1.9.5 has Optimized data format,but the gap is too large¡£
I'm a little worried about the data£¬may i ask why?
best wish
Andreas Stieger
2017-06-26 14:56:11 UTC
Permalink
 
Post by Dummy
i svnsync gzrepos(centos 6.4-svn1.6.11) to gz-mirror1(centos 7-svn1.9.5),
[...]
mirror repo(gz-mirror1) is more less than source repo(gzrepos) in sizes
the source repo(gzrepos) is 177G while the mirror one(gz-mirror1)  is only 66G
for example: revision:2012,it seems to be compressed
Possible reasons for differences in on-disk repository size:

* Target may use deduplication and intra-revision deduplication. (good)
* The user that performs the read part does not have complete repository read access
Depending on what you are trying to do, this may be bad (as the sync is incomplete.

Andreas
Nico Kadel-Garcia
2017-07-20 13:07:47 UTC
Permalink
Better deduplicaton? And did you exclude old branches with bulky binaries
in them?
Post by Dummy
i svnsync gzrepos(centos 6.4-svn1.6.11) to gz-mirror1(centos 7-svn1.9.5),
when it was done successfully, i found that the mirror repo(gz-mirror1)
is more less than source repo(gzrepos) in sizes ,the source repo(gzrepos)
is *177G* while the mirror one(gz-mirror1) is only *66G*
for example: revision:2012,it seems to be compressed
however, the mirror repo(gz-mirror1) seems OK, we can
access,checkout,show log etc.
i heard that subversion 1.9.5 has Optimized data format,but the gap is
too large。
I'm a little worried about the datamay i ask why?
best wish
Eric Johnson
2017-07-20 22:12:01 UTC
Permalink
This could be for a number of reasons. Perhaps your original repository is
an older format? If that's the case, and your mirror is a newer format,
then the newer format could be packing and finding binary duplicates much
more effectively than is possible using the older format.

Eric.
Post by Nico Kadel-Garcia
Better deduplicaton? And did you exclude old branches with bulky binaries
in them?
Post by Dummy
i svnsync gzrepos(centos 6.4-svn1.6.11) to gz-mirror1(centos 7-svn1.9.5),
when it was done successfully, i found that the mirror repo(gz-mirror1)
is more less than source repo(gzrepos) in sizes ,the source
repo(gzrepos) is *177G* while the mirror one(gz-mirror1) is only *66G*
for example: revision:2012,it seems to be compressed
however, the mirror repo(gz-mirror1) seems OK, we can
access,checkout,show log etc.
i heard that subversion 1.9.5 has Optimized data format,but the gap is
too large。
I'm a little worried about the datamay i ask why?
best wish
Loading...