我正在用bacula创建一个基于云的解决方案,可以备份我们所有的服务器。我们有很多服务器在前提上,还有一些服务器在云上,比如AWS和OPENCLOUD。我看到备份的传输速率非常低(我指的是400 each/s),对于小型FS (大约5/6GB ).这让我感到担忧,因为我们正在对这个客户端进行测试,使备份满,需要2小时才能完成,我们将在以后添加一些真正的大客户端(每个客户端大约500 of /1TB),以便对它们进行完整的备份。
这是在此客户端备份后打印的消息:
09-Nov 03:43 bacula-dir JobId 37: Start Backup JobId 37, Job=Backup-
mailserverp.2018-11-08_22.15.00_13
09-Nov 06:08 bacula-sd JobId 37: Elapsed time=02:24:57, Transfer rate=580
Bytes/second
Scheduled time: 08-Nov-2018 22:15:00
Start time: 09-Nov-2018 03:43:51
End time: 09-Nov-2018 06:08:52
Elapsed time: 2 hours 25 mins 1 sec
FD Bytes Written: 5,039,356 (5.039 MB)
SD Bytes Written: 5,048,922 (5.048 MB)
09-Nov 01:09 bacula-dir JobId 36: Start Backup JobId 36, Job=Backup-
nagios.2018-11-08_22.15.00_12
09-Nov 03:43 bacula-sd JobId 36: Elapsed time=02:34:39, Transfer rate=386.9 K
Bytes/second
Elapsed time: 2 hours 34 mins 47 secs
FD Bytes Written: 3,590,358,216 (3.590 GB)
SD Bytes Written: 3,590,441,488 (3.590 GB)
09-Nov 00:38 bacula-dir JobId 35: Start Backup JobId 35, Job=Backup-
bapuppet01.2018-11-08_21.25.00_11
09-Nov 00:38 bacula-sd JobId 34: Elapsed time=02:11:17, Transfer rate=35.68 K Bytes/second
Scheduled time: 08-Nov-2018 21:05:00
Start time: 08-Nov-2018 22:27:30
End time: 09-Nov-2018 00:38:52
Elapsed time: 2 hours 11 mins 22 secs这条消息显示了三台服务器,两台在前提上,一台在基于OpenCloud云的主机上。考虑到这种结构,我们假设onpremise将是所有备份中最慢的。但是,在使用iperf命令测试连接之后,我们会看到以下内容:
FROM SERVER TO CLIENT
[root@otc-bacula ~]# iperf -c 172.xx.xx.xxx -p 9102 -i 2 -t 60
------------------------------------------------------------
Client connecting to 172.xx.xx.xxx, TCP port 9102
TCP window size: 45.0 KByte (default)
------------------------------------------------------------
[ 3] local 172.xx.xx.xxx port 33902 connected with 172.xx.xx.xxx port 9102
[ ID] Interval Transfer Bandwidth
[ 3] 0.0- 2.0 sec 95.9 MBytes 402 Mbits/sec
[ 3] 2.0- 4.0 sec 180 MBytes 754 Mbits/sec
[ 3] 4.0- 6.0 sec 132 MBytes 554 Mbits/sec
[ 3] 6.0- 8.0 sec 70.6 MBytes 296 Mbits/sec
FROM CLIENT TO SERVER ################################
[root@v-nagios ~]# iperf -c 172.xx.xxx.xxx -p 9102 -i 2 -t 60
------------------------------------------------------------
Client connecting to 172.xx.xxx.xxx, TCP port 9102
TCP window size: 16.0 KByte (default)
------------------------------------------------------------
[ 3] local 172.xx.xxx.xxx port 41538 connected with 172.xxx.xxx.xxx port 9102
[ ID] Interval Transfer Bandwidth
[ 3] 0.0- 2.0 sec 3.75 MBytes 15.7 Mbits/sec
[ 3] 2.0- 4.0 sec 4.00 MBytes 16.8 Mbits/sec
[ 3] 4.0- 6.0 sec 2.50 MBytes 10.5 Mbits/sec
[ 3] 6.0- 8.0 sec 4.38 MBytes 18.4 Mbits/sec
[ 3] 8.0-10.0 sec 3.50 MBytes 14.7 Mbits/sec
[ 3] 10.0-12.0 sec 2.12 MBytes 8.91 Mbits/sec我已经尝试过解池选项,但我没有看到任何改进。有没有人知道测试bacula性能的方法或脚本,以及如何改进?我可以尽量减少FileSets,只是为了备份我们真正需要的东西,但是仍然有很多文件/ gygas。
这是Bacula-Dir.conf,如果有帮助的话:
Director { # define myself
Name = bacula-dir
DIRport = 9101 # where we listen for UA connections
DirAddress = 172.19.120.106
QueryFile = "/etc/bacula/query.sql"
WorkingDirectory = "/var/spool/bacula"
PidDirectory = "/var/run"
Maximum Concurrent Jobs = 10
Password = "123456" # Console password
Messages = Daemon
Heartbeat Interval = 1
}
Storage {
Name = File
# Do not use "localhost" here
Address = 172.19.120.106 # N.B. Use a fully qualified name here
SDPort = 9103
Password = "123456"
Device = FileStorage
Media Type = File
}
Pool {
Name = File
Pool Type = Backup
Label Format = OpenCloud-
Recycle = yes # Bacula can automatically recycle Volumes
AutoPrune = yes # Prune expired volumes
Volume Retention = 365 days # one year
Maximum Volume Bytes = 100G # Limit Volume size to something reasonable
Maximum Volumes = 350 # Limit number of Volumes in Pool
}我认为这个信息是有用的,但是如果你需要更多的服务信息或者需要我做些什么,我很乐意提供所需的数据.我真的需要提高这个解决方案的传输速率。
发布于 2018-11-15 12:27:37
我意识到我正在使用MySql作为DB后端,并且在插入数据的大容量上遇到了麻烦。因此,我再次进行了新的安装,但这次使用Postgresql作为数据库,传输速率几乎达到40 mb/s。大进步
我会把我自己的答案标为正确的
发布于 2019-12-04 22:17:24
下面是一些关于Bacula调优的建议:
而且,如果你还没有,这也是一个好主意,加入杆菌-用户邮件列表,在那里你可以搜索档案,并获得帮助,其他巴库拉问题。https://www.bacula.org/support/email-lists/
最后,巴库拉系统可以提供调优指导,如果你有订阅。
注:我在Bacula系统公司工作,担任CDMO。
https://stackoverflow.com/questions/53269100
复制相似问题