Feature #6676
closedAdd the number of component instances as additional resource
Added by Evgeny Novikov almost 9 years ago. Updated over 7 years ago.
100%
Description
In addition to calculation and caching of sums of wall/CPU time and maximums of memory per all invocations of a given component for all its parents it would be great to calculate and cache the number of component instances. This will add more dynamics to corresponding pages since one will be able to see how many various components already finished.
The number of component instances can be shown like the total number of safes, unsafes and unknowns.
Updated by Evgeny Novikov over 7 years ago
- Priority changed from Normal to High
Raise the issue priority since it blocks the high priority issue.
Updated by Vladimir Gratinskiy over 7 years ago
- Due date set to 06/13/2017
- Status changed from New to Resolved
- % Done changed from 0 to 100
Implemented in branch "feature_6676". You need migrations and recalculate "Cache of instances of components" for all existing jobs on manager page.
Updated by Evgeny Novikov over 7 years ago
- Status changed from Resolved to Open
Please, merge the latest master into this branch since there are nontrivial conflicts.
Updated by Vladimir Gratinskiy over 7 years ago
- Status changed from Open to Resolved
The master was merged.
Updated by Evgeny Novikov over 7 years ago
- Status changed from Resolved to Open
I suppose to merge two tables "Consumed resources" and "Instances of components". By "The number of component instances can be shown like the total number of safes, unsafes and unknowns" I assumed something like this:
Consumed resources ASE (2): 2.1 s 2.2 s 32 MB AVTG (4): 10 min 2.7 min 51 MB CPAchecker (10): 1.1 min 58 s 240 MB Core (1): 5.2 min 810 ms 29 MB EMG (10): 1.2 min 53 s 34 MB LKBCE (4): 10 min 5.5 min 33 MB LKVOG (4): 10 min 280 ms 33 MB RSB (10): 5.9 min 24 s 46 MB RSG (10): 250 ms 210 ms 31 MB SA (10): 25 s 23 s 51 MB Sub-job (4): 0 ms 0 ms 0 B TR (2): 150 ms 120 ms 30 MB VTG (4): 12 min 25 s 46 MB Weaver (10): 1.5 min 1.4 min 33 MB Total: 58 min 13 min 240 MB
Updated by Vladimir Gratinskiy over 7 years ago
Data was merged. If job is in progress then the data will be like (the first number is number of finished instances, the second - is total number of instances). The "consumed resources" help text was updated):
@ Consumed resources ASE (1/2): 2.1 s 2.2 s 32 MB AVTG (2/4): 10 min 2.7 min 51 MB CPAchecker (10/10): 1.1 min 58 s 240 MB Core (1/1): 5.2 min 810 ms 29 MB EMG (10/10): 1.2 min 53 s 34 MB LKBCE (4/4): 10 min 5.5 min 33 MB LKVOG (4/4): 10 min 280 ms 33 MB RSB (10/10): 5.9 min 24 s 46 MB RSG (10/10): 250 ms 210 ms 31 MB SA (10/10): 25 s 23 s 51 MB Sub-job (4/4): 0 ms 0 ms 0 B TR (2/2): 150 ms 120 ms 30 MB VTG (4/4): 12 min 25 s 46 MB Weaver (10/10): 1.5 min 1.4 min 33 MB Total: 58 min 13 min 240 MB @
Updated by Evgeny Novikov over 7 years ago
Vladimir Gratinskiy wrote:
Data was merged. If job is in progress then the data will be like (the first number is number of finished instances, the second - is total number of instances). The "consumed resources" help text was updated):
This a really nice additional suggestion. Observing the current behavior I thought that it would be great to show the total number of running component instances even if none of them terminated. This will result in showing "Component name (0/N) -". Otherwise one can show that some instances started but didn't terminate yet just when one of them terminated ("Component name (1/N) instance consumed resources").
Updated by Vladimir Gratinskiy over 7 years ago
Evgeny Novikov wrote:
Vladimir Gratinskiy wrote:
Data was merged. If job is in progress then the data will be like (the first number is number of finished instances, the second - is total number of instances). The "consumed resources" help text was updated):
This a really nice additional suggestion. Observing the current behavior I thought that it would be great to show the total number of running component instances even if none of them terminated. This will result in showing "Component name (0/N) -". Otherwise one can show that some instances started but didn't terminate yet just when one of them terminated ("Component name (1/N) instance consumed resources").
I didn't understand it :)
Actually I show (<Total> - <Unfinished>)/<Total> instances numbers. Report can be unfinished in 2 cases:
1) Component is running
2) The job is corrupted while component was running.
So it is usefull to detect unfinished components with this approach.
Updated by Evgeny Novikov over 7 years ago
Vladimir Gratinskiy wrote:
Evgeny Novikov wrote:
Vladimir Gratinskiy wrote:
Data was merged. If job is in progress then the data will be like (the first number is number of finished instances, the second - is total number of instances). The "consumed resources" help text was updated):
This a really nice additional suggestion. Observing the current behavior I thought that it would be great to show the total number of running component instances even if none of them terminated. This will result in showing "Component name (0/N) -". Otherwise one can show that some instances started but didn't terminate yet just when one of them terminated ("Component name (1/N) instance consumed resources").
I didn't understand it :)
Actually I show (<Total> - <Unfinished>)/<Total> instances numbers. Report can be unfinished in 2 cases:
1) Component is running
2) The job is corrupted while component was running.
So it is usefull to detect unfinished components with this approach.
The number of running/total instances and corresponding consumed resources aren't shown until at least one instance terminates and consumed resources become known. So I never can see "0/N". For instance, Core runs from the beginning, but I don't see "Core 0/N -".
Updated by Evgeny Novikov over 7 years ago
- Status changed from Open to Closed
This is exactly what I wanted. Great! I merged the branch to master in 6c6ddd7e. Don't forget to migrate databases and to recalculate caches if necessary after update.