Parent Directory
| 
Revision Log
| Links to HEAD: | (view) (annotate) |
| Sticky Revision: |
open details in new window
display currnet and total shards while bulk_load
pass sack_port as second argument to client bin/split.sh can again start two nodes on localhost
details about node memory usage
collect /proc statistics from nodes
push favicon.ico so we can distinguinsh Sack tabs in browser
add IP addresses to node's title
sort nodes in display
bulk_load all nodes in parallel (20-40Mb/s on my network)
- move all pending node status into $info->{pending} and ignore forked for refresh
- do cpio of new version *after* fork for massive speedup
add pending class to nodes only if there is anything pending
move Results to top of page (plays well with refresh)
color nodes based on their state and refresh page if something is pending (load, view)
send Refresh: header
support loading of single shard
fix warning
added checkboxes to show just some keys on graph
tweak nodes output
display shard paths as title
refresh shard allocation from view response
refactor and extend client server configuration - load shards only if they are not allready loaded - collect client responses and return it in repl using . (dot) - push all unknown repl commands to clients - eval on client now returns result to server - save will dump current chards on node to /tmp/sack.data and reload it on next startup
support multiple url args which collapse into array and use it for filter
show results <h1> only if there is reasults and include gnuplot for date, keys
sum of all merge counts doesn't make sense since it contains dumplicate keys from each node
display number of merged keys from nodes
display first shard leter and show/hide info
refactor into run_view and display view overview table
improve shards output
factor out load_shard and display map in html
color code nodes red = forked green = ping
move table generation in separate package and add sorting
load shards and run views over web
reogranize $info->{shard} into path => node
simple http listener which dumps $info
use etc/cloud as default if it exists
added clean which drops all shards from clients
merge view results on shards from clients on server
send views to clients and collect loaded shards
load shard by shard whole dataset into clients as they request it
report unknown commands from repl
extract to_all
start clients from lib/Sack/
move to package, show usage and rename fork_node to fork_ssh
begin integration of protocol v3
push whole /srv/Sack/ to nodes, find absolute client.pl path
added info command
ping all clients
deduce client path from server's
take cloud path from CLOUD
copy files using cpio, use ssh config and read cloud nodes from file
collect interesting $info and send it to repl
make server resiliant to repl disconnect
added external read-eval-loop readline interface
new client-server protocol based on IO::Select
This form allows you to request diffs between any two revisions of this file. For each of the two "sides" of the diff, enter a numeric revision.
| ViewVC Help | |
| Powered by ViewVC 1.1.26 |