Home > Uncategorized > Conversations with Fred: Thread Safe

Conversations with Fred: Thread Safe

The script is flipped for this one. I’m actually teaching Fred something.

[3/11/2009 12:23:56 PM] Matt Brown says: dfs is blazing at 313.25, but no files have been received for DFS-Encrypted
[3/11/2009 12:24:11 PM] Matt Brown says: 313.25kbps
[3/11/2009 12:24:31 PM] Frederic Boismenu says: in plain english?
[3/11/2009 12:25:16 PM] Matt Brown says: GE01 is receiving DFS traffic at 313.25Kbps, however, it has not received any files in the DFS-Encrypted DFS replicated folder since 11:30 NY time
[3/11/2009 12:26:40 PM] Frederic Boismenu says: ok…
[3/11/2009 12:26:44 PM] Frederic Boismenu says: not great
[3/11/2009 12:26:46 PM] Frederic Boismenu says: good lucj
[3/11/2009 12:27:49 PM] Matt Brown says: no sr
[3/11/2009 12:28:59 PM] Frederic Boismenu says: any clue of what’s going on?!
[3/11/2009 12:29:15 PM] Matt Brown says: well… earlier, there was a jumble
[3/11/2009 12:29:32 PM] Matt Brown says: there were WMI errors and I almost freaked out
[3/11/2009 12:29:44 PM] Matt Brown says: however, i found no errors with WMI
[3/11/2009 12:29:55 PM] Matt Brown says: I’m unsure what happened
[3/11/2009 12:30:01 PM] Matt Brown says: the backup was still going,
[3/11/2009 12:30:25 PM] Frederic Boismenu says: ?!
[3/11/2009 12:30:36 PM] Matt Brown says: there was something wrong with dfs today… however, after restarting the DFS replication service on DC01; all replicated folders except my good friend DCM-Encrypted began to replicate
[3/11/2009 12:30:57 PM] Frederic Boismenu says: . . .
[3/11/2009 12:31:00 PM] Frederic Boismenu says: I dunno
[3/11/2009 12:36:34 PM] Matt Brown says: the time in paris is the same as in geneva?
[3/11/2009 12:36:44 PM] Frederic Boismenu says: yea
[3/11/2009 12:36:58 PM] Frederic Boismenu says: i dunno if we change at the same time though
[3/11/2009 12:37:02 PM] Frederic Boismenu says: why?
[3/11/2009 12:37:04 PM] Matt Brown says: ahh okay
[3/11/2009 12:37:08 PM] Matt Brown says: i just talked to laurent
[3/11/2009 12:37:25 PM] Matt Brown says: He communicated with Colt… they will be sending someone out next week to do a loopback test
[3/11/2009 12:37:40 PM] Matt Brown says: he will let us know the final date
[3/11/2009 12:37:42 PM] Frederic Boismenu says: oui tomo
[3/11/2009 12:37:49 PM] Matt Brown says: domo
[3/11/2009 12:38:21 PM] Matt Brown says: hah
[3/11/2009 12:38:29 PM] Matt Brown says: i can see you’ve typed and erased
[3/11/2009 12:38:36 PM] Matt Brown says: it kinda drives me crazy
[3/11/2009 12:38:45 PM] Matt Brown says: hah
[3/11/2009 12:38:51 PM] Matt Brown says: so, he will let us know
[3/11/2009 12:38:57 PM] Matt Brown says: okay?
[3/11/2009 12:39:00 PM] Frederic Boismenu says: tomo = tomorrow
[3/11/2009 12:39:04 PM] Frederic Boismenu says: he’s coming tomorrow
[3/11/2009 12:39:07 PM] Matt Brown says: oh he called you?
[3/11/2009 12:39:09 PM] Matt Brown says: OH!
[3/11/2009 12:39:11 PM] Frederic Boismenu says: yea
[3/11/2009 12:39:11 PM] Matt Brown says: okay!
[3/11/2009 12:39:12 PM] Matt Brown says: hah
[3/11/2009 1:18:45 PM] Matt Brown says: hey I have good news! DFs-Encrypted is staging on DC01!
[3/11/2009 1:18:54 PM] Matt Brown says: I know you’re excited
[3/11/2009 1:19:47 PM] Frederic Boismenu says: yea right
[3/11/2009 1:20:35 PM] Frederic Boismenu says: what was the issue?
[3/11/2009 1:21:11 PM] Matt Brown says: apparently, there was an issue with WMI
[3/11/2009 1:21:17 PM] Matt Brown says: I’m looking into why it happened
[3/11/2009 1:21:26 PM] Matt Brown says: wait… no there was no issue with WMI
[3/11/2009 1:21:27 PM] Frederic Boismenu says: ok, alrigtht
[3/11/2009 1:21:30 PM] Frederic Boismenu says: ag
[3/11/2009 1:21:32 PM] Frederic Boismenu says: ah?
[3/11/2009 1:21:38 PM] Matt Brown says: something happened with the DFSR
[3/11/2009 1:21:42 PM] Matt Brown says: I don’t know what
[3/11/2009 1:21:56 PM] Frederic Boismenu says: ok, let me know what you find out plz!
[3/11/2009 1:22:00 PM] Matt Brown says: DFS was reporting that there was an issue with WMI,
[3/11/2009 1:22:09 PM] Matt Brown says: but, WMI was reporting that it was a-okay
[3/11/2009 1:22:12 PM] Matt Brown says: (y)
[3/11/2009 1:22:19 PM] Matt Brown says: so, I don’t know
[3/11/2009 1:22:44 PM] Matt Brown says: however, right now, all folders (except t:) are replicating, but DFS-Encrypted is staging
[3/11/2009 1:22:51 PM] Matt Brown says: which is a pre-replication action
[3/11/2009 1:23:01 PM] Frederic Boismenu says: yea
[3/11/2009 1:23:08 PM] Matt Brown says: I’m not sure why it’s doing that
[3/11/2009 1:23:16 PM] Matt Brown says: maybe the service restart, but it seems odd
[3/11/2009 1:23:19 PM] Frederic Boismenu says: try to understand waht’s going on
[3/11/2009 1:23:23 PM] Matt Brown says: it shouldn’t have to re-evaluate the entire state
[3/11/2009 1:23:28 PM] Matt Brown says: it shouldn’t have to re-stage files
[3/11/2009 1:23:33 PM] Matt Brown says: yea, sorry :)
[3/11/2009 1:23:33 PM] Matt Brown says: hah
[3/11/2009 1:23:35 PM] Matt Brown says: i’ll leave you alone
[3/11/2009 1:23:55 PM] Frederic Boismenu says: yea trying to read here
[3/11/2009 1:24:07 PM] Frederic Boismenu says: my cooking book
[3/11/2009 1:24:15 PM] Matt Brown says: sweet
[3/11/2009 1:24:33 PM] Matt Brown says: why was the daemon script restarting the script script so frequently?
[3/11/2009 1:24:40 PM] Matt Brown says: was it from testing?
[3/11/2009 1:45:30 PM] Frederic Boismenu says: it doesn’t restart it …
[3/11/2009 1:45:56 PM] Frederic Boismenu says: RMDS_Monitor-deamon wakes every minute to check it’s alive
[3/11/2009 1:46:10 PM] Frederic Boismenu says: before it was every 5 min…
[3/11/2009 1:46:21 PM] Frederic Boismenu says: I just shorten it to 1 min
[3/11/2009 1:47:26 PM] Frederic Boismenu says: I found a bug today…
[3/11/2009 1:48:21 PM] Frederic Boismenu says: the “subprocess” unit is actually not thread safe
[3/11/2009 1:49:47 PM] Frederic Boismenu says: oh you are refereing of the emails of this morning!
[3/11/2009 1:49:54 PM] Frederic Boismenu says: yea… I was debugging again
[3/11/2009 1:50:39 PM] Matt Brown says: cool
[3/11/2009 1:50:42 PM] Matt Brown says: what does thread safe mean?
[3/11/2009 1:51:04 PM] Frederic Boismenu says: oh yea, sorry
[3/11/2009 1:51:10 PM] Matt Brown says: Did you cron it oto 24/7?
[3/11/2009 1:51:17 PM] Matt Brown says: no just market hours?
[3/11/2009 1:51:38 PM] Frederic Boismenu says: a function is said thread safe if you can call it at the same time from different thread and still works
[3/11/2009 1:52:16 PM] Matt Brown says: ahh okay
[3/11/2009 1:52:31 PM] Matt Brown says: that’s an odd function to not be thread safe, no?
[3/11/2009 1:52:44 PM] Frederic Boismenu says: not really
[3/11/2009 1:53:13 PM] Frederic Boismenu says: some optimisation can be not thread safe
[3/11/2009 1:53:27 PM] Frederic Boismenu says: of some callback mecanism
[3/11/2009 1:53:43 PM] Frederic Boismenu says: usually it’s due to the use of a global variable
[3/11/2009 1:53:55 PM] Frederic Boismenu says: holding some kind of state data
[3/11/2009 2:07:00 PM] Matt Brown says: …sorry
[3/11/2009 2:07:08 PM] Matt Brown says: global variable holding state information?
[3/11/2009 2:07:15 PM] Matt Brown says: why would that affect thread safeness?
[3/11/2009 2:07:27 PM] Matt Brown says: oh… because the other thread may want to change that?
[3/11/2009 2:07:45 PM] Frederic Boismenu says: you got it
[3/11/2009 2:08:04 PM] Frederic Boismenu says: any share stuff and kaboom
[3/11/2009 2:13:10 PM] Matt Brown says: archived: http://mattdbrown2.blogspot.com/
[3/11/2009 2:17:29 PM] Frederic Boismenu says: k
[3/11/2009 4:22:00 PM] Matt Brown says: wow
[3/11/2009 4:22:14 PM] Matt Brown says: t:\ contains 486,875 files.
[3/11/2009 4:22:41 PM] Matt Brown says: DFS has been “staging” it since 11:30AM and it has staged 13300 files
[3/11/2009 4:22:45 PM] Matt Brown says: Camille was right
[3/11/2009 4:23:01 PM] Matt Brown says: we should reduce the size, although this situation shouldn’t have happened, it did
[3/11/2009 4:23:09 PM] Matt Brown says: and now we’re stuck with this slow reconciliation process
[3/11/2009 4:29:38 PM] Matt Brown says: 50 hours total staging time
[3/11/2009 4:29:42 PM] Matt Brown says: then the replication will happen
[3/11/2009 4:29:45 PM] Matt Brown says: awful
[9:16:54 AM] Matt Brown says: good morning fred!
[9:17:08 AM] Matt Brown says: If you have a minute, I’d like to re-test the skype video connection
[9:17:12 AM] Matt Brown says: is that even possible?
[11:11:20 AM] Frederic Boismenu says: hey Matt
[11:11:30 AM] Frederic Boismenu says: looking for some software that was on the X:\
[11:11:46 AM] Frederic Boismenu says: in something like X:\DCM\CONF\Software
[11:11:57 AM] Frederic Boismenu says: all our libraries for delphi
[11:12:25 AM] Frederic Boismenu says: where there was the Teechart
[11:16:09 AM] Matt Brown says: I:\!Devel\Delphi
[11:16:13 AM] Matt Brown says: pardon
[11:16:21 AM] Matt Brown says: \\dynamic-ge01\_install\!Devel\Delphi
[11:16:43 AM] Frederic Boismenu says: there’s nothing in there
[11:17:17 AM] Frederic Boismenu says: we had much more than that
[11:17:30 AM] Matt Brown says: I moved anything I saw in there
[11:17:34 AM] Frederic Boismenu says: including some quite expensive Intel Libraries
[11:17:35 AM] Matt Brown says: if it doesn’t exist in DCM
[11:17:44 AM] Matt Brown says: I didn’t delete anything
[11:18:18 AM] Matt Brown says: I definately didn’t migrate anything off of X:
[11:18:29 AM] Matt Brown says: but the X: file still exists, correct?
[11:18:34 AM] Frederic Boismenu says: yes
[11:18:38 AM] Matt Brown says: okay
[11:18:43 AM] Frederic Boismenu says: can you see w/ nick plz?
[11:18:53 AM] Matt Brown says: sure
[11:18:58 AM] Frederic Boismenu says: tahnks matt
[11:20:56 AM] Matt Brown says: sure
[11:20:58 AM] Matt Brown says: i’ll elt you know
[11:22:49 AM] Matt Brown says: did you all do anything with the server yesterday?
[11:23:03 AM] Matt Brown says: I got a BSOD caused by a driver issue, crashed ntoskrnl
[11:23:17 AM] Matt Brown says: pardon: ntkrnlmp.exe
[11:23:44 AM] Frederic Boismenu says: dunno what your question means…
[11:23:52 AM] Frederic Boismenu says: we worked as usual
[11:23:53 AM] Matt Brown says: Did you do anything to GE01?
[11:24:01 AM] Matt Brown says: meaning, physically, like hook up an external drive?
[11:24:06 AM] Frederic Boismenu says: no
[11:24:07 AM] Matt Brown says: or install a driver?
[11:24:07 AM] Matt Brown says: okay
[11:24:14 AM] Matt Brown says: humm… alright… thanks
[11:24:21 AM] Matt Brown says: i’ll take a look and see what happened
[11:24:26 AM] Frederic Boismenu says: we had a technician who came this morning
[11:24:26 AM] Matt Brown says: DFS is recovered
[11:24:29 AM] Matt Brown says: i tested it with jamal
[11:24:33 AM] Matt Brown says: both ways
[11:24:39 AM] Frederic Boismenu says: ok
[11:24:49 AM] Matt Brown says: the upload speed of your site is a bottleneck, however, it is functioning
[11:24:50 AM] Matt Brown says: right now
[11:24:58 AM] Frederic Boismenu says: do you have a better understand how it works, and how to optimize it?
[11:25:21 AM] Frederic Boismenu says: what are our dl/ul speed here?
[11:25:25 AM] Matt Brown says: yes… however, there is little optimization
[11:25:49 AM] Frederic Boismenu says: so what happened?
[11:25:49 AM] Matt Brown says: I’ve got down 5-20Mbps, up 500Kbps-5Mbps on paper
[11:25:56 AM] Matt Brown says: what happened with?
[11:26:03 AM] Frederic Boismenu says: DFS
[11:26:43 AM] Matt Brown says: I believe we’re running into a “race condition” with updates being performed on files on both sides, and it creates a large backlog of changes
[11:26:47 AM] Matt Brown says: and it snowballs
[11:26:53 AM] Matt Brown says: I /believe/
[11:27:00 AM] Matt Brown says: I need to keep logging and see what happens over time
[11:27:25 AM] Frederic Boismenu says: so shall we do as camille said, replicate only a smal amount of data?
[11:27:36 AM] Matt Brown says: right now, for instance, GE is receiving 22894 bytes per second from DC01
[11:27:43 AM] Frederic Boismenu says: fragment the shit in several replicated places?
[11:27:54 AM] Matt Brown says: DC01 is receiving 91 bytes per second from GE01
[11:28:01 AM] Frederic Boismenu says: ye ayea yea
[11:28:10 AM] Matt Brown says: are you speaking of reducing the staging time in the future?
[11:28:36 AM] Frederic Boismenu says: I am just asking a question
[11:28:36 AM] Matt Brown says: yes, nick is moving a lot of files tonight
[11:28:50 AM] Matt Brown says: i’m missing the sliders train
[11:28:53 AM] Matt Brown says: i’ll be back shortly
[11:29:01 AM] Frederic Boismenu says: ok
[11:29:56 AM] Matt Brown says: sorry
[11:29:57 AM] Matt Brown says: back
[11:30:02 AM] Matt Brown says: i gave them my card
[11:30:11 AM] Matt Brown says: so, do you understand where I’m at in my head about DFS?
[11:30:40 AM] Frederic Boismenu says: no
[11:30:43 AM] Matt Brown says: okay
[11:31:07 AM] Matt Brown says: I believe that a race condition caused the issues previously, where there were reportedly updates to files on both sides
[11:31:07 AM] Frederic Boismenu says: just asking you what we should do according to you w/o technical details
[11:31:17 AM] Matt Brown says: oh
[11:31:18 AM] Matt Brown says: start using DFS
[11:31:22 AM] Matt Brown says: and let me know if you see errors
[11:31:28 AM] Matt Brown says: and I’ll let you know if I see errors
[11:31:37 AM] Frederic Boismenu says: – . –
[11:31:38 AM] Matt Brown says: if possible, increase the bandwidth at your site ;)
[11:31:50 AM] Matt Brown says: that’s my non technical answer
[11:32:00 AM] Frederic Boismenu says: . . .
[11:32:04 AM] Matt Brown says: |||
[11:32:05 AM] Frederic Boismenu says: what’s the plan?
[11:32:09 AM] Matt Brown says: what plan?
[11:32:24 AM] Frederic Boismenu says: to have something that works in a better way
[11:32:34 AM] Frederic Boismenu says: (and later replicate level2 data)
[11:33:12 AM] Matt Brown says: Let me ask you: What is an acceptable latency of data?
[11:33:21 AM] Matt Brown says: 2 minutes… 1 day… 1 week?
[11:33:31 AM] Matt Brown says: 0 seconds
[11:33:32 AM] Frederic Boismenu says: more 2 minutes
[11:33:32 AM] Matt Brown says: ?
[11:33:40 AM] Matt Brown says: okay… increase bandwidth at your site
[11:33:48 AM] Frederic Boismenu says: idealy camille was thinking seconds
[11:34:14 AM] Matt Brown says: DFS is fine, we can reduce the amount in a “DFS Replicated Folder” so that staging will be reduced
[11:34:23 AM] Matt Brown says: if an delay with staging happens again
[11:35:07 AM] Matt Brown says: however, staging is DFS’s reconciliation of changes within it’s own DB… so out of the 450,000 files total, only 1450 were staged yesterday
[11:35:22 AM] Frederic Boismenu says: did you test the influence of number of files on replication?
[11:35:34 AM] Matt Brown says: there is none
[11:35:50 AM] Frederic Boismenu says: you tested it?
[11:35:59 AM] Matt Brown says: we’re live with it… what you speak?
[11:36:16 AM] Matt Brown says: we’ve got a “DFS Replicated Folder” with 450,000 files
[11:36:18 AM] Frederic Boismenu says: that we had issues
[11:36:18 AM] Matt Brown says: replicating now
[11:36:33 AM] Matt Brown says: [11:26:43 AM] Matt Brown says: I believe we’re running into a “race condition” with updates being performed on files on both sides, and it creates a large backlog of changes
[11:26:47 AM] Matt Brown says: and it snowballs
[11:26:53 AM] Matt Brown says: I /believe/
[11:27:00 AM] Matt Brown says: I need to keep logging and see what happens over time
[11:36:50 AM] Matt Brown says: I knew that woudl come in handy
[11:36:56 AM] Frederic Boismenu says: what about testing?!
[11:37:01 AM] Frederic Boismenu says: rather than waiting
[11:37:12 AM] Matt Brown says: how can I test this?
[11:37:21 AM] Frederic Boismenu says: I dunno … think about it
[11:37:29 AM] Matt Brown says: yea, i created a 10MB file, it transferred
[11:37:36 AM] Matt Brown says: Jamal created an 8MB file it transferred
[11:37:41 AM] Matt Brown says: the speeds are known
[11:37:53 AM] Matt Brown says: I made changes to the 8MB file on this side, it transferred
[11:37:59 AM] Matt Brown says: I made changes on that side, it transferred
[11:38:08 AM] Frederic Boismenu says: we need to understand the influence of the amount of data in the same replicated folder
[11:38:17 AM] Frederic Boismenu says: exemple:
[11:38:25 AM] Frederic Boismenu says: if you try to replicate the same file
[11:38:38 AM] Frederic Boismenu says: in a replicated folder holding 10000000000 files
[11:38:51 AM] Frederic Boismenu says: is the same that replicating the same file in another folder
[11:38:57 AM] Frederic Boismenu says: containing 1 file?
[11:39:09 AM] Matt Brown says: Yes.
[11:39:21 AM] Matt Brown says: after staging is completed, it is the same thing
[11:39:35 AM] Frederic Boismenu says: when staging occurs?
[11:39:58 AM] Matt Brown says: staging is for new files and changes
[11:40:42 AM] Matt Brown says: http://technet.microsoft.com/en-us/library/cc782648.aspx
[11:44:01 AM] Matt Brown says: “DFS Replication uses staging folders to act as caches for new and changed files to be replicated from sending members to receiving members.”
[11:45:15 AM] Frederic Boismenu says: blah blah
[11:45:23 AM] Frederic Boismenu says: concreatly
[11:45:44 AM] Matt Brown says: concretely what?
[11:45:57 AM] Frederic Boismenu says: what happen if we create a new file?
[11:46:08 AM] Frederic Boismenu says: will it browse all replicated files?
[11:46:18 AM] Matt Brown says: no
[11:46:37 AM] Matt Brown says: I’ll try to find the exact flow, but from my currently knowledge:
[11:46:50 AM] Frederic Boismenu says: what I am trying to understand is when we can take a hit from this staging stuff
[11:47:52 AM] Matt Brown says: Indexing service sees file> notifies DFS> DFS message sent to other member servers of replicated folder> staging takes place on “source” server (where change was made) [checksum against database + file compression algorithm]> diff sent to other member servers
[11:48:16 AM] Matt Brown says: whenever a file is changed staging will occur
[11:48:36 AM] Matt Brown says: whenever a file is created
[11:48:53 AM] Matt Brown says: whena file is deleted or moved, another DFS notification is sent to other member servers
[11:49:26 AM] Frederic Boismenu says: ok… so it’s looks good in theory?
[11:50:07 AM] Frederic Boismenu says: if we restart a server are we going to wait hours of the servers to realize they are happy?
[11:52:12 AM] Matt Brown says: of course, it’s Microsoft
[11:52:23 AM] Matt Brown says: I mean, that it looks good in theory
[11:52:31 AM] Matt Brown says: :)
[11:52:39 AM] Matt Brown says: only if staging is to take place
[11:52:46 AM] Frederic Boismenu says: but not for the restart?
[11:52:48 AM] Matt Brown says: if a lot of changes occur during the “down time”
[11:53:13 AM] Matt Brown says: can you ask camille to skype back in
[11:56:54 AM] Frederic Boismenu says: is everything ok? Camille’s speaking on skype now?
[11:57:16 AM] Matt Brown says: hey
[11:57:17 AM] Matt Brown says: sorry
[11:57:21 AM] Matt Brown says: so, yes.
[11:58:53 AM] Matt Brown says: hold on… let me regain focus
[11:58:58 AM] Frederic Boismenu says: matt could you put the polycom
[11:59:03 AM] Frederic Boismenu says: in the middle of the table
[11:59:08 AM] Matt Brown says: it’s not long enoug
[11:59:09 AM] Matt Brown says: h
[11:59:25 AM] Matt Brown says: if there’s a sound issue, we can use the phone
[11:59:38 AM] Frederic Boismenu says: camille wants it the closest possible
[11:59:42 AM] Matt Brown says: or get a usb extension cable
[11:59:47 AM] Matt Brown says: it is
[11:59:49 AM] Frederic Boismenu says: you have one?
[11:59:52 AM] Matt Brown says: we don’t
[11:59:55 AM] Frederic Boismenu says: . . .
[11:59:57 AM] Frederic Boismenu says: ok
[12:00:05 PM] Matt Brown says: in the future, you know
[12:00:09 PM] Matt Brown says: we can fix these things
[12:00:20 PM] Matt Brown says: however, right now, if sound is an issue, we can use the phone
[12:00:30 PM] Frederic Boismenu says: camille says it’s on the computer and that you can move it…
[12:00:40 PM] Frederic Boismenu says: (middle man is a sucky job)
[12:05:15 PM] Matt Brown says: okay
[12:05:16 PM] Matt Brown says: all set
[12:05:22 PM] Matt Brown says: so, do you understand DFS?
[12:05:44 PM] Frederic Boismenu says: kinda
[12:05:51 PM] Frederic Boismenu says: understand what we discuss
[12:06:20 PM] Frederic Boismenu says: except I am not confortable w/ the consequences of restarting of the services or computer
[12:06:34 PM] Matt Brown says: it does a self-check when the service restarts
[12:06:35 PM] Frederic Boismenu says: even in the best case (no modifications)
[12:06:51 PM] Matt Brown says: it goes and chats with the other members “what’s up? What’s the deal with these here Replicated Folders”
[12:07:03 PM] Matt Brown says: then they reply
[12:07:34 PM] Matt Brown says: For instance, when teh service restarted the other day, DCM, _Install, Home, Share were fine
[12:07:38 PM] Matt Brown says: DFS-Encrypted was staging
[12:07:44 PM] Frederic Boismenu says: but is it browsing all the files to make sure there’s not local modifications for exemple?
[12:07:55 PM] Matt Brown says: No files with stage unless changes are made
[12:08:12 PM] Matt Brown says: if the service is down, the reconciliation process will see what fiels need to be staged, then the staging will occur
[12:08:17 PM] Frederic Boismenu says: how does it know if it must stage it or not?
[12:08:19 PM] Matt Brown says: once staging completes, replication will take palce
[12:08:28 PM] Matt Brown says: checked against internal DB
[12:08:35 PM] Matt Brown says: keeps a GUID of files
[12:08:42 PM] Matt Brown says: and checksums and stuff
[12:08:50 PM] Frederic Boismenu says: so it need to check all files for modifications…
[12:09:04 PM] Frederic Boismenu says: and checksums all replicated files again?
[12:09:10 PM] Matt Brown says: in a JetDB, for which the limitations are the theorhetical limitations of DFS-Replication
[12:09:24 PM] Frederic Boismenu says: yea whatever
[12:09:31 PM] Matt Brown says: It will perform the diff-algo on files that changed
[12:09:35 PM] Matt Brown says: and repl the diffs
[12:10:13 PM] Matt Brown says: [now i understand what it must be like for me to be asking you all these ludacious questions]
[12:10:17 PM] Matt Brown says: ;)
[12:10:17 PM] Frederic Boismenu says: so this phase run length is function of the number and size of files, correct?
[12:10:28 PM] Frederic Boismenu says: lol, oui
[12:10:44 PM] Matt Brown says: What phase… man I’m busting out Visio
[12:10:57 PM] Matt Brown says: i understand your wording, but what phase?
[12:11:02 PM] Frederic Boismenu says: ok my understanding is:
[12:11:16 PM] Frederic Boismenu says: 1. check file modifications
[12:11:23 PM] Frederic Boismenu says: 2. stage modfications
[12:11:29 PM] Frederic Boismenu says: 3. etc
[12:11:39 PM] Frederic Boismenu says: I am concerned about phase 1
[12:11:44 PM] Frederic Boismenu says: when it starts
[12:11:58 PM] Matt Brown says: service off> files change> service on> DFS looks at files> Files changed? > stage file changes> Files staged? > repl files>
[12:12:20 PM] Frederic Boismenu says: my 1. = your “DFS looks at files”
[12:12:26 PM] Frederic Boismenu says: this is a really not casual thing
[12:12:30 PM] Matt Brown says: you’re concerned with checking the checksums of the files, or the staging process
[12:12:44 PM] Frederic Boismenu says: my 1. = your “DFS looks at files”
[12:13:01 PM] Matt Brown says: DFS Replication detects changes on the volume by monitoring the update sequence number (USN) journal, and DFS Replication replicates changes only after the file is closed.
[12:13:04 PM] Matt Brown says: http://technet.microsoft.com/en-us/library/cc781091.aspx
[12:13:11 PM] Matt Brown says: that’s a distraction
[12:13:14 PM] Matt Brown says: sorry, let me look more
[12:13:34 PM] Matt Brown says: “DFS Replication uses many sophisticated processes to keep data synchronized on multiple servers.” (rofl)
[12:13:38 PM] Matt Brown says: there’s your answer
[12:13:40 PM] Matt Brown says: :D
[12:13:43 PM] Frederic Boismenu says: lol
[12:14:01 PM] Frederic Boismenu says: I say … restart DC01 and look at the disk activity ;)
[12:14:29 PM] Matt Brown says: that’s fine
[12:14:38 PM] Matt Brown says: actually: http://technet.microsoft.com/en-us/library/cc781091.aspx
[12:14:44 PM] Matt Brown says: explains what happens well
[12:14:51 PM] Matt Brown says: conflict resolution heuristic
[12:14:53 PM] Matt Brown says: USB journal watcher
[12:15:00 PM] Matt Brown says: “vector exchange protocol”
[12:15:48 PM] Matt Brown says: http://technet.microsoft.com/en-us/library/cc779936.aspx
[12:15:56 PM] Matt Brown says: say not the storage server team blog
[12:16:06 PM] Frederic Boismenu says: headache
[12:16:16 PM] Matt Brown says: nice
[12:18:17 PM] Matt Brown says: so anyway… i’ll try to find you an answer on how it performs the checksums and stuff
[12:18:20 PM] Matt Brown says: where it’s stored
[12:18:29 PM] Matt Brown says: will this satisfy your hunger for knowledge?
[12:19:44 PM] Frederic Boismenu says: yes for now
[12:20:01 PM] Matt Brown says: okay
[12:20:19 PM] Matt Brown says: i wonder if microsoft employees are able to use any microsoft software for free at work
[12:20:24 PM] Frederic Boismenu says: let me know when you have more info on the checksum shit (like practical one would be good)
[12:20:40 PM] Matt Brown says: a practical answer ?
[12:20:43 PM] Matt Brown says: NEVER!
[12:20:52 PM] Frederic Boismenu says: like ask for a MS project?
[12:21:10 PM] Matt Brown says: i mean, i wonder if they can just install anything they want at anytime
[12:21:21 PM] Matt Brown says: like a free for all on the entire software bundles
[12:21:32 PM] Matt Brown says: or if they do licensing per department or some shit
[12:21:46 PM] Matt Brown says: i should write raymond chen an email
[12:21:49 PM] Matt Brown says: the old new thing guy
[12:21:54 PM] Matt Brown says: hah
[12:21:58 PM] Matt Brown says: i gotta eat
[12:22:11 PM] Matt Brown says: when i find out abotu the checksum and DB I’ll let you know.

Advertisements
  1. No comments yet.
  1. No trackbacks yet.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: