F

FileMap

Settings | Report Duplicate

1

I Use This!

Activity Not Available

Commits : Listings

Analyzed about 1 year ago. based on code collected about 1 year ago.

Commit Message	Contributor	Files Modified	Lines Added	Lines Removed	Code Location	Date
Jan 18, 2023 — Jan 18, 2024 Showing page 10 of 21 Search / Filter on:
fm split can now take a field range in addition to an individual field	Mike Fisk	More...				over 12 years ago
show actual nodename rather than qname	Mike Fisk	More...				over 12 years ago
regression fix	Mike Fisk	More...				over 12 years ago
skip inactive nodes in a few more places and ensure when picking that at least 1 active node is returned	Mike Fisk	More...				over 12 years ago
promote a setting that we need even for inactive nodes	Mike Fisk	More...				over 12 years ago
Improved retries logic and reporting. Default to 0 retries except for all2all operations which now default to 7 with a configurable setting.	Mike Fisk	More...				over 12 years ago
better way to handle singletons than last patch --- this way we can have non-numeric lists if its just for inactive nodes. For example: [1a,3b,5c] inactive=True	Mike Fisk	More...				over 12 years ago
bugfix: didn't handle singletons correctly	Mike Fisk	More...				over 12 years ago
add example SSH replacement for SLURM. Show how to use ranges to easily define a whole cluster.	Mike Fisk	More...				over 12 years ago
moving 'df' to a LocalCommand --- but still just verbatim strings right now	Mike Fisk	More...				over 12 years ago
allow nodenames to include lists of forms such as: [1-2] (1-2) [1-2,6-9,10] (1-2,6-9,10) and automatically mark nodes inactive if we're running under SLURM and didn't get them allocated to us	Mike Fisk	More...				over 12 years ago
better way to invalidate errored-out outputs from the past. Do a dry run through that just invalidates at the very beginning. Don't invalidate after that to prevent loops where we invalidate and regenerate errors again.	Mike Fisk	More...				over 12 years ago
fix rsync bug	Mike Fisk	More...				over 12 years ago
show total input size processed	Mike Fisk	More...				over 12 years ago
remove headers from df results	Mike Fisk	More...				over 12 years ago
fm plot: remove & mark idle time intervals (which could otherwise dominate the plot and make it useless)	Mike Fisk	More...				almost 13 years ago
allow pipelines that start with a reduce or redistribute operation	Mike Fisk	More...				almost 13 years ago
try to warn on empty inputs	Mike Fisk	More...				almost 13 years ago
more robust stdin processing	Mike Fisk	More...				almost 13 years ago
fix reduce processing to handle partitions more efficiently (and probably correctly)	Mike Fisk	More...				almost 13 years ago
try to make plots start right at beginning of processing	Mike Fisk	More...				almost 13 years ago
rollback some optimizations on input globbing (cachedinputs) since they caused problems. Upstream parents can create new files and invalidate and re-create old ones, so we have to keep re-evaluating potential inputs aggressively (like we used to). file modification events would be the best way to handle this.	Mike Fisk	More...				almost 13 years ago
Rework put/store logic to be more approachable. Probably fixed some bugs in the process (hopefully more than I introduced)	Mike Fisk	More...				almost 13 years ago
ignore files that don't end in "db". Better cleanup on error. Default to 300 partitions since sqlite insists on opening a journal and consuming 3 fd's partition. This way we fit in a 1024 fd default max.	Mike Fisk	More...				almost 13 years ago
bugfixes	Mike Fisk	More...				almost 13 years ago
Plotting improvements: Speed-up timeline packing in plotting. Make horizontal boundaries tighter. Minimum height of a block is 30% (up from an invisible 0%). Remove markers at edges of bars.	Mike Fisk	More...				almost 13 years ago
color each job differently (from a finite pallete)	Mike Fisk	More...				almost 13 years ago
rewrite in python and add partition.	Mike Fisk	More...				almost 13 years ago
minor improvements	Mike Fisk	More...				almost 13 years ago
"db" is a tool used for managing a flow of flat-file databases (e.g. sqlite) in filemap jobs	Mike Fisk	More...				almost 13 years ago

←
1
2
…
6
7
8
9
10
11
12
13
14
…
20
21
→