CrashPlan restart loop

[overkill]

So on 8/11 one of my drive decided to crap out. Apparently when I pulled the bad drive out, the whole host restarts, and so did the jails and especially crashplan.
When crashplan starts up, it start analyzing my data. I didn't think much of it and leave it there. Till today, I thought "how come the GUI went away by itself, I surely didn't close it".
Turns out crashplan service has been restarting itself every so often.

And every time it restarts it make a log file which reads like this
root@crashplan_2:/usr/pbi/crashplan-amd64/share/crashplan/bin # cat restart.2016-08-17_17.44.26.log
Wed Aug 17 17:44:27 PDT 2016 : /usr/pbi/crashplan-amd64/share/crashplan/bin/restart.sh
Wed Aug 17 17:44:27 PDT 2016 : APP_BASENAME=CrashPlan
Wed Aug 17 17:44:27 PDT 2016 : DIR_BASENAME=crashplan
Wed Aug 17 17:44:27 PDT 2016 : Stopping using ./CrashPlanEngine...
Stopping CrashPlan Engine ... OK
Wed Aug 17 17:44:37 PDT 2016 : Sleeping 10 seconds...
Wed Aug 17 17:44:47 PDT 2016 : Starting using ./CrashPlanEngine...
Starting CrashPlan Engine ... Using standard startup
OK
Wed Aug 17 17:44:47 PDT 2016 : New Service Process below:
Wed Aug 17 17:44:47 PDT 2016 : Exiting restart script

My crashplan install is fairly standard, aside from setting the dedup setting (http://networkrockstar.ca/2013/09/speeding-up-crashplan-backups/) to 1
I had my Xmsx amount for both SRV and GUI to 1024. Upon noticing the problem I raised it to 2048, and eventually then 3072MB, none appear to help. Since the crashplan plugin runs on a 32-bit Java engine, I'm avoiding setting 4096MB to the memory amount.
On another note, my whole backup set is ~20TB or so. I'm hoping to finish the rest of 4.9TB before updating anything else (crashplan, freenas, CIFS flags, etc...)

Anyone seen this problem before and able to fix it?

 

[dlavigne]

Were you able to resolve this?

 

[SweetAndLow]

Crashplan is pretty broken and tough to get working on freenas these days. Most of us have just started to run it in a VM and mount our storage.

Sent from my Nexus 5X using Tapatalk

 

[overkill]

Were you able to resolve this?

Yes I was. Sort of. After increasing the amount of memory, it restarted a couple more times and then it's been stable. Weird thing is that I didn't need this much memory allocated to it before it was restarted. I'm afraid with larger data sets it'll need to be continuously run, otherwise when it's restarted there's no more memory to increase to (32bit limitation)

Crashplan is pretty broken and tough to get working on freenas these days. Most of us have just started to run it in a VM and mount our storage.

Sent from my Nexus 5X using Tapatalk

Couldn't agree more about that. Unfortunately I don't have the resources to make VM at the moment, in the future I will probably do that too. Any advice on that part, like what OS should I use on the VM? The usual choise would probably be windows but I'm thinking of Samba's performance (single threaded), I cringe a little. Maybe do a ubuntu vm and run it through NFS?

 

[SweetAndLow]

Linux VM and use whatever sharing protocol you use normally. Cifs is fine and don't worry about the cifs single threaded nonsense, it doesn't matter when just doing home networking.

Sent from my Nexus 5X using Tapatalk

 

[overkill]

Linux VM and use whatever sharing protocol you use normally. Cifs is fine and don't worry about the cifs single threaded nonsense, it doesn't matter when just doing home networking.

Sent from my Nexus 5X using Tapatalk

Does having either LACP on 4Gbps ports or a single 10Gb for local connection count as home networking? I don't have either of those yet just as I don't have resources for VM because I'm using old hardware that I got for free, but eventually I'd setup FreeNAS as a VM in ESXi. I'm already using a switch capable of LACP (that I got for free), the question would be whether I'd get one of these http://www.ebay.com/itm/Quanta-LB4M...698?pt=LH_DefaultDomain_0&hash=item418ddf2502 for the 10Gb

Cifs will probably be fine for crashplan alone but I'm trying to plan long term, probably 10 years out or so. I'll have an array comprised entirely of SSDs for things that require high performance, say... whatever I'm working on in lightroom. So fast network + fast drives, naturally I'm concerned of cifs' possible performance neck.