Failed to in-place upgrade cluster from 1.6.5.4 to 1.7.1-5

Farshid Ghods added a comment - 14/Aug/11 6:36 PM Integrated in github-ep-engine-1-7-1 #141 (See [ http://qa.hq.northscale.net/job/github-ep-engine-1-7-1/141/ ])      MB-4188 Need to add erlang path for escript command Bin Cui : Files : * management/mbdbupgrade

Thuan Nguyen added a comment - 15/Aug/11 4:54 PM Log file from broken node 165

Thuan Nguyen added a comment - 15/Aug/11 5:14 PM Log files from second cluster (10.1.6.160, 10.1.6.161 and 10.1.6.162) with repeat error "Membase Server has started on web port 8091 on node ' ns_1@10.1.6.162 '."

Thuan Nguyen added a comment - 15/Aug/11 5:25 PM I can reproduce this bug in build 1.7.1.1r-1 and log files attached

Farshid Ghods added a comment - 16/Aug/11 11:59 AM Problem signature:   Problem Event Name: APPCRASH   Application Name: mbdbconvert.exe   Application Version: 0.0.0.0   Application Timestamp: 4e2b26dd   Fault Module Name: ntdll.dll   Fault Module Version: 6.0.6002.18005   Fault Module Timestamp: 49e0421d   Exception Code: c0000026   Exception Offset: 00000000000b8fb8   OS Version: 6.0.6002.2.2.0.274.10   Locale ID: 1033   Additional Information 1: 5a33   Additional Information 2: c2d97aff665064deeb04310c84cabe59   Additional Information 3: efa7   Additional Information 4: 212b643616a2b83b8b8ab989da92ecb2 Read our privacy statement:    http://go.microsoft.com/fwlink/?linkid=50163&clcid=0x0409  

Farshid Ghods added a comment - 16/Aug/11 12:01 PM Automatic mode: running without interactive questions or confirmations. Upgrading your Membase Server to 1.7.1.1r-2-gfb59e71. The upgrade process might take awhile. Analysing... Previous config.dat file is C:\Users\ADMINI~1\AppData\Local\Temp\2\backup-8-16-2 011-09-40-39\config\ns_1\config.dat Target node: ns_1@10.1.6.38 Membase should not be running.   Please use the Control Panel to stop the Membase service. Database dir: c:/Program Files/Membase/Server/data/ns_1 Buckets to upgrade: default Checking disk space available for buckets in directory:   c:/Program Files/Membase/Server/data/ns_1   Free disk bucket space wanted: 16001024.0   Free disk bucket space available: 3870449664   Free disk space factor: 2.0   Ok. Analysis complete. Copying C:\Users\ADMINI~1\AppData\Local\Temp\2\backup-8-16-2011-09-40-39\config\ ns_1\config.dat     cp C:\Users\ADMINI~1\AppData\Local\Temp\2\backup-8-16-2011-09-40-39\config\n s_1\config.dat C:\Program Files\Membase\Server\bin\..\var\lib\membase\config\con fig.dat Ensuring bucket data directories. Ensuring bucket data directory: c:/Program Files/Membase/Server/data/ns_1/defaul t-data     mkdir -p c:/Program Files/Membase/Server/data/ns_1/default-data Upgrading buckets. Upgrading bucket: default     mbdbupgrade.exe c:/Program Files/Membase/Server/data/ns_1/default c:/Program  Files/Membase/Server/data/ns_1/default-data mbdbupgrade pid: 1624 Source version from "c:/Program Files/Membase/Server/data/ns_1/default" is 2 mbdbconvert.exe --src-pattern=%d/%b-%i.mb --src-strategy=multiMTVBDB c:/Program Files/Membase/Server/data/ns_1/default c:/Program Files/Membase/Server/data/ns_1 /default-data\default --report-every=2000 --init-file=..\etc\membase\init.sql src = multiMTVBDB@c :/Program Files/Membase/Server/data/ns_1/default dest = multiMTVBDB@c :/Program Files/Membase/Server/data/ns_1/default-data\defaul t Each . represents 2000 items moved.

Thuan Nguyen added a comment - 16/Aug/11 2:01 PM I got the same issue in unix env (centos 5.4 64bit). The cluster of 3 nodes broken when upgrade from 1.6.5.4 to 1.7.1.1r-2. Attachment are config.dat of 3 nodes 178, 179 and 180.

Aleksey Kondratenko added a comment - 16/Aug/11 3:41 PM and I was able to reproduce. Looks like config.dat is initial config on all nodes. Windows case only differs by nodes_wanted in config.dat. But still differs.