I'm new to DSS and I'm trying to replace a handmade iSCSI SAN setup with DSS, and I'm running into various problems.
iSCSI failover is correctly configured I believe: the secondary node takes resources when the primary goes down.
The first problem is: we have to shutdown iSCSI failover to add an iSCSI target to the failover. That means everything connected will either hang or crash each time we add a target.
That is a design flaw being worked on as I read. I believe I can do nothing on my end about that.
The second problem is: the web interface is painfully slow or even timeouts when "too many" (is that even possible ?) are created.
I have currently 50 iSCSI volumes and targets, and it takes several minutes to list them.
Modifying a volume (add/remove replication) via the webinterface takes the same amount of time.
Hardware is the same on both nodes:
- dual xeon quadcore
- 16 gigabytes of RAM
- areca 1680
- 8x 1terabyte HDD, RAID5 with no spare (7 terabytes effective)
- 4 gigabit network, one dedicated to replication.
Did I miss something that could lead to such slowness ?
The adding of additional volume replication tasks for the Auto Failover is not a design flaw but a feature we are currently adding to the Auto Failover feature and engineers are working to have this available in future releases but no ETA as of yet.
On the web interface being slow was this from the very begining? Also what version are you runing for the DSS V6?
Yes, the web interface has been slow since the beginning. Especially with actions on the volumes (list,create,modify,delete,etc).
But it has become unusable at this time: more than 2 minutes to list the LV on the VG, same time for every action.
It's weird to note that creating a LV with web interface takes 2 minutes but doing so with API takes 10 seconds.
As we're evaluating DSS, it's DSS V6 trial version.
For it to be slow from the beginning then something else is happening and we would need to look at the logs to see if there is some hardware issue. Send them to pre-sales@open-e.com
I sent the logs from both SANs. Primary with 50 volumes and painfully slow and secondary with 5 nodes (not all were added because the primary was unusable).
Firefox and Chrome (both latest with cleared cache) takes around 2 minutes to simply display the 50 volumes in the only volume group.
Any action on those volumes (add,edit,remove) taks the same time.
Internet Explorer is faster with around 1 minute to do the same actions. This is still very slow.
Still, API command volume_status takes around 4 seconds to display the same data.
Other API commands like volume_replication and create_iscsilv are also a lot faster (a couple of seconds instead of a couple of minutes).
Slow web interface, up to timeouts.
Reply With Quote
