innobackupex process died: many tables created during incremental backup

[Nickolay Ihalainen]

(for i in `seq 100000 200000` ; do echo "create table t$i (id integer)engine=innodb;insert into t$i values($i);" ; done ) | mysql test
#wait for finishing, create a full backup.
innobackupex /data/backups/base

(for i in `seq 200000 300000` ; do echo "create table t$i (id integer)engine=innodb;insert into t$i values($i);" ; done ) | mysql test
# and start incremental backup in parallel
nohup innobackupex --incremental /data/backups/inc/ --incremental-basedir /data/backups/base/2011-09-23_19-02-56 2>&1 | tee -a /var/log/backup.log

during the file copy time innobackupex died:
110924 00:50:36 innobackupex: Starting to backup .frm, .MRG, .MYD, .MYI,
innobackupex: .TRG, .TRN, .ARM, .ARZ, .CSM, .CSV and .opt files in
innobackupex: subdirectories of '/var/lib/mysql'
innobackupex: Backing up files '/var/lib/mysql/mysql/*.{frm,MYD,MYI,MRG,TRG,TRN,ARM,ARZ,CSM,CSV,opt,par}' (72 files)
innobackupex: Error: Broken pipe at /usr/bin/innobackupex line 336.

but xtrabackup hangs in suspended state and could be finished only by deleting xtrabackup_suspended file.

ulimit -n 1024 is used, but I have catch the same situation in production with ulimit -n 999999 and 300k tables.

[Alexey Kopytov (akopytov)]

This request is identical to the one in bug #1135441. Closing this one as a duplicate and not vice versa, because bug #1135441 is more detailed and has some discussion on potential ways to fix.