TestingServerHardware
|
⇤ ← Revision 1 as of 2005-10-31 22:27:51
Size: 2292
Comment:
|
Size: 3326
Comment:
|
| Deletions are marked like this. | Additions are marked like this. |
| Line 17: | Line 17: |
| build mode for the server livefs which the server tester just pops in, answers a few questions (which hardware does the system really have, vs. what the system thinks it has), email the result to us => done. | Create a mode for the server livefs which the server tester just pops in, answers a few questions (which hardware does the system really have, vs. what the system thinks it has), email the result to us => done. |
| Line 20: | Line 20: |
| - basic hardware recognition - Userspace tools for hardware configuration (raid controllers, etc) - Hot plug systems for blades (CPU, memory) - performance (memory bandwidth, disk, CPU, whatever) - include database workload or equivalent for high-level measurements - Burn-in, long term work load testing for stability - multi-system option for network testing (throughput, make sure that the network adapter doesn't go belly-up under load) |
* basic hardware recognition * Userspace tools for hardware configuration (raid controllers, etc) * Hot plug systems for blades (CPU, memory) * performance (memory bandwidth, disk, CPU, whatever) * include database workload or equivalent for high-level measurements * Burn-in, long term work load testing for stability * multi-system option for network testing (throughput, make sure that the network adapter doesn't go belly-up under load) |
| Line 37: | Line 37: |
Vendors can supply kit to us, but they will find it much easier to ship to one location. If they ship to the one and only Canonical location, then we would have to ship to whoever is doing the testing. This would cost time, money and effort and would not be the best use of our resources. If we had a central location to ship to, and a central location to test in, this would be most efficient. However, as we are a distributed organisation, this would be difficut to achieve. Unless a team is defined to be located with the test centre. It is possible to pay for this service, but we probably do not want to do that .... as we can actually charge the h/w vendors for the fact that we are certifying their h/w. But - as we would need the systems to recreate the problems that may be reported on these systems, we should plan on keeping them ... as most of the major vendors would expect . So, possible thought is having a test centre next to / nearby the support centre ......lots of things to thnk about ...... |
|
| Line 52: | Line 60: |
| # Create server testing scripts # Find a place/people to do centralized server certification (this should be near the "support center" which might need this hardware later to replicate a user issue") # Create a way to track community server test results (talk to ogra about intigration with hwdb) |
* Create server testing scripts * Find a place/people to do centralized server certification (this should be near the "support center" which might need this hardware later to replicate a user issue") * Create a way to track community server test results (talk to ogra about intigration with hwdb) |
Launchpad Entry: https://launchpad.net/distros/ubuntu/+spec/foo
Created: Date(2005-10-31T22:27:51Z) by MarkRamm
Contributors: MarkRamm
Packages affected:
Summary
We need to do a better job of testing server hardware. To do this we need to:
- Create an easy way to support and encourage community server testing
- Get up-to date hardware from IBM / HP / Sun / Apple / Dell to certify against Ubuntu Server Dapper
- Create a comprehensive server test suite
Code
Create a mode for the server livefs which the server tester just pops in, answers a few questions (which hardware does the system really have, vs. what the system thinks it has), email the result to us => done.
Test Suite Should Contain:
- basic hardware recognition
- Userspace tools for hardware configuration (raid controllers, etc)
- Hot plug systems for blades (CPU, memory)
- performance (memory bandwidth, disk, CPU, whatever)
- include database workload or equivalent for high-level measurements
- Burn-in, long term work load testing for stability
- multi-system option for network testing (throughput, make sure that the network adapter doesn't go belly-up under load)
Data preservation and migration
None
Outstanding issues
It is difficult to reconcile Canonical's globally distributed development environment with Hardware Vendor's desire to ship stuff to a single location for testing purposes.
BoF agenda and discussion
Vendors can supply kit to us, but they will find it much easier to ship to one location. If they ship to the one and only Canonical location, then we would have to ship to whoever is doing the testing. This would cost time, money and effort and would not be the best use of our resources.
If we had a central location to ship to, and a central location to test in, this would be most efficient. However, as we are a distributed organisation, this would be difficut to achieve. Unless a team is defined to be located with the test centre.
It is possible to pay for this service, but we probably do not want to do that .... as we can actually charge the h/w vendors for the fact that we are certifying their h/w.
But - as we would need the systems to recreate the problems that may be reported on these systems, we should plan on keeping them ... as most of the major vendors would expect . So, possible thought is having a test centre next to / nearby the support centre ......lots of things to thnk about ......
Rationale
Use cases
Scope
We would like to certify on 25 machines from the above companies in the Dapper timeframe, and we would like to have community testing of as many server configurations as possible. We need a way to record and track user reports that is seprate from the way we do official server certification.
Design
Implementation
There are three projects here.
- Create server testing scripts
- Find a place/people to do centralized server certification (this should be near the "support center" which might need this hardware later to replicate a user issue")
- Create a way to track community server test results (talk to ogra about intigration with hwdb)
TestingServerHardware (last edited 2008-08-06 16:30:16 by localhost)