6,954 reputation
21738
bio website
location Boston, MA
age
visits member for 5 years, 5 months
seen Nov 11 at 7:54

Oct
11
comment 50um vs. 62.5um fiber compatability
It is much much worse than "they're not remotely compatible". They're almost compatible. If you have a run 62.5->50, you will get a link and ping will work and you'll go home thinking you've set things up. Then, later, when you run more than a trivial amount of traffic over the link, things will be terribly slow because you'll have enormous packet loss in one but not the other direction. This failure mode is far worse than "it doesn't work"
Aug
23
comment tool for splitting pcap files by TCP connection?
The files that result from tcpflow are not pcaps, they are the actual tcp payloads of the tcp streams.
Jul
10
comment Expose a mongodb streaming data server that's behind a firewall, to the internet
I agree an SSH tunnel is the way to go here. Keep in mind that using an SSH tunnel will probably get the poster fired if anyone ever figures out what they're up to.
Jul
1
comment Gigabit uplink versus regular switch port
TCP expects to drop packets. Dropped packets are how TCP calibrates available bandwidth -- it increases the transmit rate until it observes dropped frames, then it backs off, then increases transmit rate until it sees dropped frames, etc. Generally on switches the "uplink" port is just another port that is identical to the others, even if it has an SFP physical link as well as a standard rj45 connector.
Jun
27
comment is zfs snapshot -r of several pools atomic?
That's really cool! Thanks for the info!
Apr
28
comment ZFS snapshots and atomic updates
@syneticon-dj -- that's not actually the behavior I observed -- I saw quite clearly that a static (in the source pool) file would not cause IO errors in the destination pool; files changing frequently on the source pool, however, would show this behavior on the target pool.
Apr
24
comment ZFS snapshots and atomic updates
And -- I could see lots of situations where I'd want to use a filessytem this way -- let's say I had a bunch of web files I wanted to atomically update on a bunch of remote servers. I could use rsync and change a symbolic link or I could send a new snapshot to all of them.
Apr
24
comment ZFS snapshots and atomic updates
I wasn't sure what would happen. Unix has a pretty convenient way of dealing with files as they change; I wasn't sure if the same conventions would apply in this scenario. If I'm reading a file and someone else changes it, I'll either see the changes or not, depending on my current offset in the file and where the change took place. Snapshots certainly add another dimension to this problem.
Apr
24
comment ZFS snapshots and atomic updates
The error is seen by the program reading the file, not the programs involved in the snapshot replication.
Aug
29
comment replace a 'space' char in filename with an underscore
This looks like the safest, most portable approach to me. Magic syntax stuff is always a little sketchy and hard for someone else to understand what's going on; this answer uses one of the 2 obvious choices for "character substitution" -- the other being tr IE var=$(echo "$var" | tr ' ' _)
Aug
29
comment replace a 'space' char in filename with an underscore
So this is way off in the weeds but you need to quote both "$1" and "${i// /_}". Also, that ${var///} stuff looks like a non-posix shell extension (bash perhaps)?
Aug
20
comment OSPF routers (with BIRD on debian) recognize each other as neighbors but can't ping each other
Also, please take the time to put the text of the screen shots into the question.
Aug
20
comment OSPF routers (with BIRD on debian) recognize each other as neighbors but can't ping each other
Router A talks to Router B over the 192.168.250.1/30 subnet (A=192.168.250.1 and B=192.168.250.2)
Jul
23
comment What happens if I delete lost+found
It isn't that they couldn't create the lost+found -- it is that it is a bad idea to create files / directories on a filesystem that's already screwed up. Instead, you just prebuild a directory that's already big enough to store the directory entries of whatever munched inodes you find in a screwed-up filesystem when you're trying to clean it up.
Jun
28
comment Recursive move utility on Unix?
cp is terrible at this sort of thing because it typically destroys hard links / soft links and replaces them with individual files. Similarly it can't be trusted to keep ACLs straight and sometimes gets confused about sparse files.
Jun
17
comment Can you set inode quotas in ZFS?
That's too bad. I work in an environment with hundreds of thousands of users each with unique UIDs and simply running find ~homedir | wc -l would take a couple months. If I had quotas it would be simpler to track these users down. As it is, I just wait until a user with 20gajillion files causes trouble for some of our other systems (malware scanning, backups, etc). Thanks!
Apr
30
comment Any options out there for kerberized ssh client on windows?
It looks like gssapi has been added to putty. I had used a clunky hacked version that would hang for 30 seconds or more if you had gssapi turned on on the client but not the remote system. I don't currently work in an environment that uses kerberos so I can't tell if that issue is the case with the stock putty version. Nevertheless, I think mainline is probably the place to start here. Thx!
Apr
24
comment New office cabling: should we use CAT7 or CAT6a?
The install isn't only the components -- it is trivial to have a cat5e install made up of entirely components rated as cat6a. The only way to tell the difference is with one of these: datacomtools.com/catalog/cat-6-tester.htm -- and not the $1300 one, you have to use the $10,000 one. And when there's fishy behavior on your "cat6a" network 3 years after it is installed, you think any typical shop will have one of these in their cable test kit?
Apr
24
comment New office cabling: should we use CAT7 or CAT6a?
I think you misunderstand -- the "Standard" only works if you've tested the whole cable install. Port to port. If your vendor doesn't do a full test, or if you substitute a non-cat6a cable as a patch cable, or if the structured cabling gets roughhoused because someone has to replace a switch -- you'll easily wind up with a non-cat6a install that still works at 10gb, but not always, such as maybe when you've plugged your 30th 10gb node in, long after the incident that actually degraded the cable.
Apr
22
comment Multiple DNS, priority of Servers
You really should specify which client you're talking about. They all behave differently, and often the behavior is documented.