[pve-devel] The "bug" in pve reflecting ZFSplugin

[Alexandre DERUMIER]

Hi,

My 2 cents, but didn't we have add recently
"update_vm_async: new asynchronous API"

https://git.proxmox.com/?p=qemu-server.git;a=blobdiff;f=PVE/API2/Qemu.pm;h=b99045e7066f6bdc7521906c56f85ab4ed530c3f;hp=ecad91534ed15e8d5e103e4c91ae679edd7b8950;hb=5555edea95d07f78b7281703270f185f5b9f105a;hpb=d8b916fdb018e2b15618ba27b6f091b9959e6ed6

+# POST/PUT {vmid}/config implementation
+#
+# The original API used PUT (idempotent) an we assumed that all operations
+# are fast. But it turned out that almost any configuration change can
+# involve hot-plug actions, or disk alloc/free. Such actions can take long
+# time to complete and have side effects (not idempotent).
+#
+# The new implementation uses POST and forks a worker process. We added 
+# a new option 'background_delay'. If specified we wait up to
+# 'background_delay' second for the worker task to complete. It returns null 
+# if the task is finished within that time, else we return the UPID.
+ 


to manage slow storage operations ?

----- Mail original ----- 

De: "Dietmar Maurer" <dietmar at proxmox.com> 
À: "Michael Rasmussen" <mir at datanom.net>, pve-devel at pve.proxmox.com 
Envoyé: Mercredi 21 Août 2013 20:31:24 
Objet: Re: [pve-devel] The "bug" in pve reflecting ZFSplugin 

> I think I am about to circle in the problem regarding ZFSplugin. To me it 
> seems like the gui makes a request to the backend and then sits waiting for 
> a response for a distinct time and if no response is given within than distinct 
> time the request will be resent for a given number of times. If a response is 
> returned within that time frame everything is ok but if no response is given 
> within that time frame a code 500 (communication error) is raised. 
> 
> Since running a remote request over SSH obviously will take longer than a 
> running it local I was wondering if the timeout the gui is using before 
> making a new request in some way could be increased by say 10 secs either 
> global or per request? Increasing the timeout 10 secs needed for especially 
> the request for creating a new volume (image) since this requires at least 4 
> successive remote requests. 

Honestly, I am not willing to add anything that add such long delays. 

If ssh does not work (a simply request should respond within millisecond), you need to find another solution. 

_______________________________________________ 
pve-devel mailing list 
pve-devel at pve.proxmox.com 
http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel