Reputation
458
Top tag
Next privilege 500 Rep.
Access review queues
Badges
1 4 15
Impact
~35k people reached

  • 0 posts edited
  • 3 helpful flags
  • 654 votes cast
May
1
comment DNS not working with Ipv6
Nslookup uses it's own libraries, dig uses the default resolver - you should be good unless you need nslookup to be able to look it up for whatever reason. Nslookup is an old tool, and while it isn't officially deprecated, most people prefer dig when available. If you are using nslookup in a script or something for monitoring, consider using dig instead.
Feb
18
comment Setting up Infiniband network - wrong route
ipoib requires standard arp - you should see ipoib entries in the arp table that will have the GUID and IP of the ib devices on the same network.
Feb
17
comment Setting up Infiniband network - wrong route
192.168.2.107 and 192.168.2.108 are on the same network - you shouldn't need a route to talk. Are they arping properly?
Oct
7
comment Storage Network Stops Transmitting Traffic
The chelsio t320 driver has an issue where it suspends the card instead of reinitializing when it hits a fatal error. The root cause is really a hardware failure. Can you see any errors in logs (fatal errors/parity errors) relating to the cxgb interface?
May
15
comment QoS Degrading VOIP
The biggest problem with VoIP is that you can't control the whole path. QoS only goes as far as your router and then the packets do as they will along the rest of the journey to their intended destination. I did a pretty thorough write-up on SU: superuser.com/questions/80580/how-to-improve-voip-sound-quality/…
May
15
comment Proper way to create an NS record in Windows DNS
As a point of being practical, I agree with you. If it works, it works. However, you don't really answer the question. If it doesn't matter, why? what exactly does Windows Server do internally to create that link? If you see my update, it appears to generate and return an A record, even if it doesn't show up in the A record list in the management console. What I'm unclear about is if that sort of "shortcut" would pose an issue under any circumstances, or if it is perfectly valid and OK to do.
Feb
27
comment Measuring network performance
What sort of control would you have over the servers? There are several options such as iperf (which is what speedtest.net uses) and SNMP. There are also better methods of using ping to look for not only latency, but jitter as well.
Feb
3
comment Scalable NAS: on-premise and minimum of 6 TB
glusterfs is a good cheap way to set up a scalable distributed file system. There is no support and it will take some effort to build it all initially, but would do what you need.
Feb
3
comment Options for SATA ports on motherboard for Files server
Much of the solution also isn't just in how you build it, but how you utilize it as well. There are a lot of storage solutions capable of managing 300TB, but depending on how you'll access it, type of reads/writes, amount of clients connecting, size of files, etc, the solution you want will vary greatly.
Oct
21
comment Transfer between two NFS shares on the same machine
why not take a pcap or use netstat to look at traffic/connections to see what, if anything, is going out over the wire? Or use strace to see what calls are being made with the mv?
Sep
30
comment Reboot a server and sign into a program
AutoIt or AutoHotKey might do the trick.
Sep
27
comment How to achieve multiple NFS/TCP connections to the same server?
see section 5.6 tldp.org/HOWTO/NFS-HOWTO/performance.html - it talks about increasing the number of nfsd instances.
Sep
27
comment How to achieve multiple NFS/TCP connections to the same server?
Multiple connections is going to be the only way to do what you're looking for. A single stream should max out around 350MB/s per thread on a fine-tuned nfs server. In any case, if we don't know what distro or nfs version you're using, it is hard to give a good recommendation.
Sep
27
comment How to achieve multiple NFS/TCP connections to the same server?
What distro are you using on the server? How much RAM? number of cores? How many disks? type of disks? average iops per disk that you are seeing? operations in queue on the disks? time in queue per op on the disk? version of nfs? number of clients connecting to the server? There is a lot that goes into perf tuning and it is important to step back and look at the big picture first, and then narrow down. If you have done this, it helps to post numbers and stats you have already gathered. Even the clients may have something to do with it (eg, buffer size, mount options, dirty pages, etc).
Sep
27
comment Network Patch Panel Connections
I have to agree with Grant. I used to cable up sites frequently and most tend to like to do it all at once for several reasons. It is cheaper if you are contracting to have it all done on the same trip. It is much easier to route cables without creating tangles and ugliness if switch groups are all routed at the same time. As far as reasons to do it the other way, if you label everything, it is easier to label end to end as you're putting them in, and if you don't know end to end when you install, it can be a pain to put ptags or ptouch flags on cable ends in a tied bundle later on.
Jan
29
comment Intermittent NFS lockups on Isilon cluster
maybe modify the min/max rpc threads?
Jan
29
comment Fastest way to delete millions of files on an Isilon IQ 72NL SAN?
disabling the journal on an Isilon is a very, very, very bad idea and will likely cause data loss across the cluster. Please do not ever attempt this.
Oct
12
comment Fastest way to delete millions of files on an Isilon IQ 72NL SAN?
A locally run rm is a bad idea on an isilon cluster. It is single threaded and will make use of only one node's power. If you want a faster method, you'd be better off with the TreeDelete job as it is multi-threaded and will use the entire cluster's compute strength to complete.
Oct
19
comment Will a router allow network traffic from a static IP?
Are we talking public or private IPs here? Can you take control of the statics in your router and just route the combined traffic to the data center gateway or do you need to just add DHCP functionality aside from what is statically assigned from the data center?
Sep
30
comment solution to store 10TB /Month
I want to expand on this. Not only can you lose a node or disk (depending on setup, even multiple nodes), the cluster rebuilds the data on its own so that you don't need to rush a new node or disk in, unlike traditional raid storage, where you require the replacement to rebuild.