Closed Bug 737982 Opened 13 years ago Closed 13 years ago

Full SSH access to a seamicro for benchmarking a Django application

Categories

(Infrastructure & Operations Graveyard :: WebOps: Other, task)

x86
macOS
task
Not set
normal

Tracking

(Not tracked)

RESOLVED WONTFIX

People

(Reporter: peterbe, Unassigned)

Details

(a shot in the dark because I've never asked for this) I'm launching two standard Django apps on seamicros (pto-dev.allizom.org and sheriff-dev.allizom.org) and there they are really really slow. Like mouse-clicking-noticibly slow. Neither sites are super fast but on a struggling vagrant box on my MBP it can push out 40 requests/sec with a concurrency of 10. With full access to a seamicro I can debug what's being so slow. In particular I suspect the I/O to the mysql is the bottleneck (as opposed to CPU or memory limits) but I can't know for sure. Can this be arranged? Can you perhaps "clone" the sheriff-dev.allizom.org VM? I intend to run a bunch of measurements of the app itself (django-debug-toolbar) and some benchmarks to see if I can tune the mod_wsgi settings or something.
Please file IT bugs using the IT bug form.
Status: NEW → RESOLVED
Closed: 13 years ago
Resolution: --- → INVALID
Filed in the wrong place, moving
Assignee: nobody → server-ops
Status: RESOLVED → REOPENED
Component: Operations → Server Operations
Product: Mozilla Services → mozilla.org
QA Contact: operations → phong
Resolution: INVALID → ---
Version: unspecified → other
Component: Server Operations → Server Operations: Web Operations
QA Contact: phong → cshields
I can't grant a node with such access alongside others running production (infrasec) However, most of the stuff will be leaving the atom seamicros soon enough anyway.
Status: REOPENED → RESOLVED
Closed: 13 years ago13 years ago
Resolution: --- → WONTFIX
(In reply to Corey Shields [:cshields] from comment #3) > I can't grant a node with such access alongside others running production > (infrasec) > No problem. > However, most of the stuff will be leaving the atom seamicros soon enough > anyway. Didn't know that. However, now that I've done a bunch of stresstests and benchmark of this app locally (or rather, in a debian vagrant) I'll be more than happy to run the same stuff on a VM later so we can debug potential bottlenecks in a typical Playdoh app. Hit me up when time (and hardware) allows.
Component: Server Operations: Web Operations → WebOps: Other
Product: mozilla.org → Infrastructure & Operations
Product: Infrastructure & Operations → Infrastructure & Operations Graveyard
You need to log in before you can comment on or make changes to this bug.