Forum
Welcome, Guest
Username: Password: Remember me

TOPIC:

Volume size limitation 5 years 1 month ago #1797

  • Salvatore Costantino
  • Salvatore Costantino's Avatar
  • Offline
  • Posts: 722
FYI
Multiple multiple IPs is OK for the HA, but, be aware that they must ALL be reachable to return true.

Please Log in or Create an account to join the conversation.

Volume size limitation 5 years 1 month ago #1798

  • Giorgio Catena
  • Giorgio Catena's Avatar Topic Author
  • Offline
  • Posts: 30
what do you mean? I want to drive a switch only when effectively the other host plus the defined witnesses are not reachable. Do you mean that the fallback would take blace only when all the defined witnesses ip are reachable again ?

Please Log in or Create an account to join the conversation.

Volume size limitation 5 years 3 weeks ago #1804

  • Salvatore Costantino
  • Salvatore Costantino's Avatar
  • Offline
  • Posts: 722
Correct. The logic for the list of heuristic ip addresses (witness IPs) is AND, meaning all of the IPs in the list mush be reached to count as a single vote.

Please Log in or Create an account to join the conversation.

Volume size limitation 5 years 3 weeks ago #1805

  • Giorgio Catena
  • Giorgio Catena's Avatar Topic Author
  • Offline
  • Posts: 30
Thanks a lot

Please Log in or Create an account to join the conversation.

Volume size limitation 5 years 3 weeks ago #1807

  • Giorgio Catena
  • Giorgio Catena's Avatar Topic Author
  • Offline
  • Posts: 30
Hi Salvatore,
at the end I was able to compare my ideas with a techinician for our surveilllance systems (that is the aim of the cluster I would like to build).
He confirm me that as a worst case we can work with a datarate (in terms of written bit) of 300Mbit/sec). This incluse already a little overhed for cams that has a lot of movement.
Due to this, considering what you told me about the overhead given by nosan and replica, I should think that the hardware must be able to give 390MBit/sec (30% overhead).
The idea was to build a Raid 10 configuration with 18 SAS 7.2 K RPM disks controlled obviously by a battery backed cache controller (Controller LH modulare HPE Smart Array P408i-a SR Gen10).
Apart the data rate they rise me a doubt about the presence of a sql db and the fact that the machine on the slave nodes if the HA switch will have or not a consistent db once it is turned on.

kind regards

Giorgio

Please Log in or Create an account to join the conversation.

Volume size limitation 5 years 3 weeks ago #1808

  • Salvatore Costantino
  • Salvatore Costantino's Avatar
  • Offline
  • Posts: 722
You should achieve roughly the same performance as DAS for any VM running on the master. You can easily measure this yourself by running throughput test from dom0 and domu running on the master. The slave will be somewhat degraded, but still adequate for any workloads that are not diskbound.

With that said, below are some crude test result from about 5 years ago. These were done HP G5 servers with 10K SAS in a RAID 1+0

From a VM running on the master backed by iscsi-ha
[root@test1 tmp]# dd if=/dev/zero of=testf bs=10000 count=10000
10000+0 records in
10000+0 records out
100000000 bytes (100 MB) copied, 0.81666 s, 122 MB/s ( 976Mbps)

From same VM as above when running on slave backed by iscsi-ha
[root@test1 tmp]# dd if=/dev/zero of=testf bs=10000 count=10000
10000+0 records in
10000+0 records out
100000000 bytes (100 MB) copied, 1.34623 s, 74.3 MB/s (594.4Mbps)

Regarding data integrity, by default we run DRBD in synchronous mode using protocol C which provides for data integrity between the primary and secondary nodes. To further eliminate the possibility of a DB write not being committed, you should look into disabling any write caches (which could slow performance a bit). See below post for the exact DB scenario you describe and the config change required.

halizard.com/forum/suggestion-box/80-missing-data-on-ha#446

Please Log in or Create an account to join the conversation.