Description of problem: After complete geo-rep data sync, slave has more files than master. There were some node reboots. But there were no deletes or renames done of master volume's data set. Version-Release number of selected component (if applicable): glusterfs-3.4.0.44rhs-1.el6rhs.x86_64 How reproducible: Not sure. Was able to reproduce only twice. Once with glusterfs-3.4.0.43rhs and now with glusterfs-3.4.0.44rhs-1.el6rhs.x86_64 Steps to Reproduce: 1. Create and start a geo-rep session between using the tar+ssh method from 2*2 dist-rep master and 2*2 dist-rep slave. 2. Now keep rebooting a single node many times while the session is in progress and the data is being created. Use smallfiles_cli.py and crefi.py for data creation. 3. After the data creation is over, set the checkpoint. Actual results: After checkpoint is completed, this was the arequal-checksum on master and slave. [root@lightning master]# cat /tmp/areequal-checksum-master egular files : 457616 Directories : 2062 Symbolic links : 2940 Other : 0 Total : 462618 Metadata checksums Regular files : 92e9 Directories : 3e9 Symbolic links : 3e9 Other : 3e9 Checksums Regular files : 7d4ed1e031bee8bf2610acbb012a3a32 Directories : 306120322c56586f Symbolic links : a047250611b353f Other : 0 [root@lightning master]# cat /tmp/areequal-checksum-slave Entry counts Regular files : 457618 Directories : 2062 Symbolic links : 2940 Other : 0 Total : 462620 Metadata checksums Regular files : 92e9 Directories : 3e9 Symbolic links : 3e9 Other : 3e9 Checksums Regular files : 7d4ed1e031bee8bf2610acbb012a3a32 Directories : 5866601c533a3d1d Symbolic links : a047250611b353f Other : 0 Total : 93c6f1702b5daaf There are two more regular files in slave. [root@lightning master]# diff /mnt/master/find-master /tmp/find-slave 60990a60991 > ./level00/5283d7b8~~IVNU5550EX 211516a211518 > ./level03/52842dd6~~2D6F4SGUR2 These were the gfid of the files in slave [root@lightning slave]# getfattr -d -n glusterfs.gfid -e hex /mnt/slave/level00/5283d7b8~~IVNU5550EX getfattr: Removing leading '/' from absolute path names # file: mnt/slave/level00/5283d7b8~~IVNU5550EX glusterfs.gfid=0x6f7b88f83bf64db7a01ae54f8a45cec5 [root@lightning slave]# getfattr -d -n glusterfs.gfid -e hex /mnt/slave/level03/52842dd6~~2D6F4SGUR2 getfattr: Removing leading '/' from absolute path names # file: mnt/slave/level03/52842dd6~~2D6F4SGUR2 glusterfs.gfid=0x3dd0abbdedab4ef58299c4ecd35e96fc [root@spitfire ]# gluster v geo master falcon::slave status detail MASTER NODE MASTER VOL MASTER BRICK SLAVE STATUS CHECKPOINT STATUS CRAWL STATUS FILES SYNCD FILES PENDING BYTES PENDING DELETES PENDING FILES SKIPPED ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- spitfire.blr.redhat.com master /rhs/bricks/brick0 falcon::slave Active checkpoint as of 2013-11-14 16:25:54 is completed at 2013-11-14 16:26:05 Changelog Crawl 428809 0 0 0 0 harrier.blr.redhat.com master /rhs/bricks/brick2 hornet::slave Active checkpoint as of 2013-11-14 16:25:54 is completed at 2013-11-14 21:44:26 Changelog Crawl 274809 0 0 0 191 mustang.blr.redhat.com master /rhs/bricks/brick1 interceptor::slave Passive N/A N/A 0 0 0 0 0 typhoon.blr.redhat.com master /rhs/bricks/brick3 lightning::slave Passive N/A N/A 385350 0 0 0 0 And on master the files were not present at all. Expected results: Slave should not have more files than master. They should be exactly in sync. Additional info: I will archive all the files from master and slave. Possibly the gfid's of all the files in master and slave as well.
Closing this bug since RHGS 2.1 release reached EOL. Required bugs are cloned to RHGS 3.1. Please re-open this issue if found again.