Feature requests

[lohelle]

I have a couple of feature requests for HA setups:

1: sync order (what devices to sync first)

2: number of simultaneous syncs

Now it looks like it is random what target is synced when (full and fast sync).
When the storage is not saturating the sync links, multiple simultaneous syncs could speed up "getting back to normal" a lot after a reboot/crash.

[Anatoly (staff)]

No problem!
Before I`ll put it on the list I want to make sure that I understand the purpose of improvements. So, could you please clarify what in your opinion should be achieved by implementing both of that?
Thank you very much for the input!

[lohelle]

Sync order: Lets say we have production and dev/test on the same server. It would be nice to sync production targets first.

As to the simultaneous syncs, we have 10Gig sync links, and we cannot saturate the sync link during sync with many of our targets. This is limited by storage performance (2 x SSD in RAID1). If we could sync 2-3 simultaneous, we could cut resync time in half at least.

[Vladislav (Staff)]

Hi lohelle,

1. Thanks a lot for this request. We already has it on the way, but this is not a priority. Now we have +1 for this feature.
2. Simultaneous sync is already available. Sync can be start manually for any device. But, nevertheless, it is +1 to the auto simultaneous sync feature.

Thank you for your contribution.

[pstoianov]

1. SNMP monitoring - OIDs for performance, amount of connections per target, node's status;

2. SNMP traps for at least MAJOR and CRITICAL events such as node disconnections and sync issues;

3. L2 cache status, usage, utilization and performance;

4. L1 cache usage and performance;

5. Counter of flash-based disk write cycles with threshold setup capability;

6. Automatic check and promotion in case all nodes were restarted;

7. Rsyslog support;

8. Triming in case whole target is flash-based;

9. CLI interface to manage nodes;

10. Option that Metadata is placed on flash-based disks;

11. Configuration backup by one-click or CLI;

12. HOT option to enable/disable L1 or/and L2 cache;

13. HOT option to change WT<=>WB L2 cache mode;

14. HOT L2 cache size resizing;

15. Storage Utilization space display;

16. Performance display: IO size + Total IO + Throughput;

17. Configuration import with automatic disk creation for quick node rebuild;

18. Snapshot jobs per schedule;

I'll not add for now public and private cloud features possible....

[Vladislav (Staff)]

Wow

Thanks a lot for these requests.

[danisoto]

6. Automatic check and promotion in case all nodes were restarted;

12. HOT option to enable/disable L1 or/and L2 cache;

13. HOT option to change WT<=>WB L2 cache mode;

+1

These are must have for high-grade production!

* Automatic poweron in case of failure of all nodes is a request done for years!

* CLI control of the state of caches is a requirement for safe control based on UPS status: when UPS is working with batteries then all the caches need to flush and the WT mode must be activated. With scripting this can be done, but current CLI don't have the option for FLUSH, WT<=>WB, etc.

I hope this will be implemented soon.

[anton (staff)]

Guys I don't reply anything but it does not mean I don't care We're gathering feedback and after Microsoft Ignite and next big update we'll sit in our war room and do a brainstorming what to add. So be ready to answer our annoying questions THANK YOU!!! WE APPRECIATE YOUR HELP!!!