22000+ entries in 0.05s
mod6: To disgorge a heap profile snapshot at a given moment:
mod6: your email just says this:
mod6: is that flag a deal-breaker?
mod6: yup, i'm using the same flags exactly. except I need to add the -datadir flag.
mod6: i can do another, sec.
mod6: there was just one line in that file: P=(ID=11738 N=(./bitcoind) T=0.008000)
mod6: yeah, i did that in my last paste: ./bitcoind -datadir=/mnt/btc-dev/.bitcoin profileheapdump v0531-heapdump-002.txt
mod6: ah i see what you mean. ok.
mod6: is that a flag for igprof?
mod6: im not sure what you mean exactly?
mod6: ok so that doesn't matter?
mod6: yeah, the strange part is that im missing that `igprof.mp.gz`
mod6: i'll put together a rundown
mod6: hmm. this time it didn't spit me out a `igprof.mp.gz`. the log file seems to be growing slowly though.
mod6: ok, started back up... looking ...
mod6: ok. i'll try again :]
mod6: (19:37) <+mod6> it was executed like so: `nohup $INSTAREA/bin/igprof -d -mp -z -o igprof.mp.gz ./bitcoind -datadir=/mnt/btc-dev/.bitcoin -daemon -profileheap >& igtest.mp.log &`
mod6: my data dir isn't under /root/.bitcoin so every time I invoke bitcoind i need to tell it where to look for its stuff.
mod6: i can show that, gimme a sec.
mod6: i gotta just have one silly thing out of place is my guess, but i can't tell what it is yet.
mod6: im doing this like this:
mod6: sorry, i should have thrown that in
mod6: P=(ID=11559 N=(./bitcoind) T=0.008000)
mod6: <+ascii_modem> gotta ask it with rpc cmd to take each dumpshot << when I take the dump it just puts this one line in there...
mod6: <+ascii_modem> read the ml post again << ok just re-read for the 3rd time. im /pretty/ sure my execution command (first line of that last dpaste ^^) looks correct. can you confirm?
mod6: ok. what did i miss?
mod6: nothing is echo'd out to the terminal.
mod6: that's all i have. output proper?
mod6: do you got time for you to get a full breakdown from me? i can put one together quick.
mod6: I've started it up, with a command (pasted above); however, I'm not 100% sure if it's running properly. the size of the log doesn't seem to be growing at all, and not much is dumped when i run: ./bitcoind -datadir=/mnt/btc-dev/.bitcoin profileheapdump v0531_heapdump_003.txt ( for example )
mod6: hi ascii_modem. I've got igprof built, and i tested it with the test code in the COMPILE.txt file. i've got your igprof_hooks patch applied to the v0.5.3.1-RELEASE source, and i've built it dynamically.
mod6: it was executed like so: `nohup $INSTAREA/bin/igprof -d -mp -z -o igprof.mp.gz ./bitcoind -datadir=/mnt/btc-dev/.bitcoin -daemon -profileheap >& igtest.mp.log &`
mod6: asciilifeform: check this out, does this look ok so far? ^^
mod6: now to try to fire this thing up.
mod6: alright, bitcoind build cleanly with the igprof_hooks patch applied.
mod6: as as separate note, I will not be running NMON with this full-sync test -- I want to get as clean of a igprof profile as I can. Don't want any thing else to disturb it's collection.
mod6: Unless anyone objects in the next 30 minutes or so.
mod6: After a profile with vanilla v0.5.3.1-RELEASE, I'll further profile with the two Orphanage Patches.
mod6: [ I'm running igprof with v0.5.3.1-RELEASE first as we know with this version the OOM-KILL signal is recieved. Maybe we can get a deeper glimpse to what's going on here. ]
mod6: ok asciilifeform's igprof_hooks patch applied to v0.5.3.1-RELEASE
mod6: ok igprof is built... now to just run it with v0.5.3.1-RELEASE...
mod6: naw, i think it's interesting. let us know how that ends up.
mod6: too many different things I gotta get done first.
☟︎ mod6: you're gonna get a pogo going? i still need to get a chance to get mine fired up.
mod6: I'm gonna work on trying to get IgProf going here now.
mod6: Thanks to trinque for the help.
mod6: Will update again as I have them.
mod6: I'm gonna try a few other things, might need to get this figured out on another machine. We'll see.
mod6: Update on gentoo build on physical box: so trinque & I worked on trying to get this Grub situtation figured out for quite a while lastnight. Seems that maybe something with grub2 doesn't like my hardware. Grub 0.9x is no longer supported... so I dunno.
mod6: look how rusted the rocker arms are
mod6: However, this could be tabled until later; when we add in a previously discussed #b-a certified seed discovery mechanism.
mod6: yeah, alf submitted it back in feb i think.
mod6: Also, i believe that the removal of the dnsseed is going to be on the docket as well. but again, we'll see how it goes.
mod6: ok back. yeah, so... I'll dig into this for sure here this month. Hopefully sooner, rather than later. I wanna make sure the whole thing 'fits in head' before we move forward. But so far, looking good.
mod6: yeah, like i said, i gotta look into it. :]
mod6: B is recieved before A, so B is 'orphan'
mod6: so [A_tx]<-[B_tx] are both sent
mod6: more over, is there a "normal" situation that arises where a guy is trying to create a transaction and it becomes an orphan? his system time is off? many things i suppose...
mod6: <+asciilifeform> i.e. what causes it to fall into that category ? << right, some sort of malformed tx or one that has some weird time in the tx that makes it get rejected or some such thing.
mod6: asciilifeform: ok. i'll have to think on that a bit more. i wanna be sure of what these changes imply. anyway, a guy can always patch his own if he wishes in the mean time.
mod6: sure. i was kinda wondering about the converse side where there is an orphaned TX, and then what? guy just resends?
mod6: ben_vulpes: should read up on this too ^^
mod6: yeah, i think just in here, we can discuss. be better for me to spend a bit of time thinking about it for a day or two.. sometime this week maybe.
mod6: But so far, I feel good that these two patches fully sync'd and with some reasonable results.
mod6: <+asciilifeform> mircea_popescu: and must nitpick, but it isn't the current therealbitcoin yet. not until patches are ratified. << we'll get there. I'd like to have a discussion at some point about the possible negitive effects of at least the TX Orphanage Amputation patch.
mod6: yup, will get to it... hopefully :] lot of things to work though this month.
mod6: im gonna try some profiles this month.
mod6: asciilifeform: yeah, i agree. it's not great, but it's something.
mod6: i'll continue to do more testing. we'll hopefully get something out of it.
mod6: no, it's system level. all I can tell you is all that is running as far as 3rd party things are: bitcoind & nmon, otherwise i've had `vmstat 1` going.
mod6: memory usage graph looks very similar to the previous run with the OrphanageThermonuke graph. That's expected, but just sayin'. Lookin good.
mod6: wow! yeah, this one didn't hit the OOM-Kill either, ran good the whole time without problems.