WinkyWolly's questions

I have a exceptionally simple setup I'm playing with however I'm banging my head against the wall as to why it's not working.

Right now I have two Vyatta routers (R1, R2) connected locally. Both are connected via a cross connect. I've setup a BGP session between the two - that's working as intended. There is no route-maps / policies in place. I'm simply trying to advertise a single route from R1 to R2 via BGP (I'm not using any IGP yet, merely a static route).

Everything is seemingly working as it's supposed to - R1 and R2 have a BGP session open and R2 see's the routes R1 is attempting to advertise:

vyatta@R2# run show ip bgp neighbors 10.0.50.10 received-routes
BGP table version is 5, local router ID is 10.0.60.10
Status codes: s suppressed, d damped, h history, * valid, > best, i - internal
Origin codes: i - IGP, e - EGP, ? - incomplete

Network          Next Hop            Metric LocPrf Weight Path
*> 10.0.2.0/24      10.0.50.10                             0 1 i

Total number of prefixes 1

Now the problem is that's just routes it's seeing however it's not actively adding it to the routing table. Running show ip bgp will list nothing as well as show ip bgp neighbor 10.0.50.10 routes. R2 can access the next hop (R1) at 10.0.50.10.

Am I missing something terribly obvious?

I know ZFS is usually used for numerous reasons including it's amazing mirroring features. However in this instance the machine I've been tasked with setting up has a RAID-1 setup so I only have one logical disk.

My reason for wanting ZFS is primarily that of snapshots - is this the right tool for the job or is there a better way to efficiently emulate ZFS snapshot?

I'm having a chicken or egg problem - is it possible to install a base RHEL system with one disk (/dev/sda) using ZFS? It seems all references I can find use it to pool numerous disks with the base system not being mentioned. How would I achieve this?

Please let me know if I can clarify this at all as I'm fully aware my explanation is lackluster at best.

My colocation provider has setup HSRP for us and has provided us 2 drops (plugged into 1/1/1 and 1/1/2) from their routers as well as a /29 with 2 HSRP IP's and the rest of the /29 usable for the switch.

I know that HSRP should be pretty much plug and play as long as the ports are in the same VLAN.

However I'm confused as to how to the IP setup should work.

Currently I've assigned port 1/1/1 the full /29 however I think this is incorrect because if I were to unplug it port 1/1/2 wouldn't work as it has no IP assignment.

What's the proper way to do this?

I'm new to networking and am working on setting up a simple flat private VLAN setup with a single Brocade FCX switch, a few hosts and connectivity from the facility.

Unfortunately due to my lack of some basic understanding I'm failing to grasp a small issue. Here is the current setup:

VLAN 100 (Primary VLAN) - Has a VE assigned 192.168.5.1/24 - pvlan mapping for 101, 102 to 1/1/10

VLAN 101 (Community VLAN) - Has two hosts

VLAN 102 (Isolated VLAN) - Has one host

Now the PVLAN itself functions properly. If I hook up a server/PC to 1/1/10 I can see all of the broadcast traffic from the PVLAN fine. I'm however somewhat clueless as to how to get this to route / "work" as intended.

Where would I map the Primary VLAN to? Would I point it to the access port (1/1/1)? Is there a way to get it to use the VE device I created?

In all of the examples I'm seeing it's showing the Primary VLAN hooked up to a firewall (which makes sense, just not in my environment) however I'm looking to literally send it upstream directly.

Here is an example config:

vlan 100 by port
untagged ethe 1/1/10 
 router-interface ve 1
 pvlan type primary
 pvlan mapping 101 ethe 1/1/10
 pvlan mapping 102 ethe 1/1/10
!
vlan 101 by port
 untagged ethe 1/1/9 eth 1/1/12                                        
 pvlan type community
!
vlan 102 by port
 untagged ethe 1/1/11 
 pvlan type isolated

Thank you for any feedback / direction :).

I'm starting to learn basic switching / networking and I've hit a small hiccup in understanding a fundamental thing. In the scenario I have a L3 switch (Brocade in my environment) and the facility gave me a /24 (We'll say 10.1.0.0/24).

Now would you assign the full 10.1.0.0/24 range to switchport 1? Or would you assign say a /30 to switchport 1 and subnet the remainder depending on your needs (VLAN's, default VLAN, single port)?

The former makes no sense to me as there would be overlapping subnets if you attempted to assign anything within that subnet elsewhere.

What's the proper procedure for this?

I'm attempting to setup a mapper device prior to boot as it requires use of my /var partition which is relied heavily upon startup.

My issue is it's seemingly failing and not providing any output - I'm unsure if this is something I'm doing terribly wrong or a limitation I'm simply not aware of.

I'm using flashcache and have it loading with my initrd. I also have the binaries for flashcache working fine it however just fails.

I've merely added the following to the initrd init script:

<snip>
setuproot
echo Creating flashcache volumes for var
flashcache_create -p around sdcachehome /dev/sdb1 /dev/sda7
echo Switching to new root and running init.
switchroot
</snip>

Unfortunately it's spitting out a unhelpful "failed" and then continues on with the boot process.

Creating flashcache volumes for var
cachedev sdcachehome, ssd_devname /dev/sdb1, disk_devname /dev/sda7 cache mode WRITE_AROUND
block_size 8, cache_size 0
Flashcache metadata will use 38MB of your 15995MB main memory
echo 0 756613242 flashcache /dev/sda7 /dev/sdb1 sdcachehome 3 2 8 0 473949641118 72 8 | dmsetup create sdcachehome failed
Switching to new root and running init.

Is this a viable way to be doing this? Can I setup a mapper device in this way?

Thanks!

I would like to have FlashCache cache my /var partition however I can't seem to get it to play nice upon bootup (IE: not really sure how to do it).

I'm not sure if I need to modify the initramfs/use DKMS or if I can do it in user-land during bootup. The issue I'm running into is /var mounts early and therefore the device is busy (whether it by generally by syslogd).

I'm positive this can be resolved by modifying the initramfs although I simply haven't fiddled with it enough to get it working.

They have instructions on how to boot your root partition however I'm not sure if these instructions would apply to my use case.

Any help / pointers in the right direction would be absolutely splendid.

I'm looking to compile a custom kernel from the official source tree (kernel.org). The machine itself runs the latest stable RHEL kernel without issues but I can't seem to figure out why it's panicking on any stock kernel. I suspect it's a driver issue however I'm relatively new at debugging this.

The kernel attempted to compile with a stock config is 2.6.32.55. I've also attempted to compile it with the previous working kernel config to no avail.

The machines specs are as follows:

• Motherboard: SuperMicro X9SCI-LN4F Intel Xeon SingleProc SATA
• CPU: Intel Xeon-SandyBridge E3-1270-Quadcore
• RAID-1 with an Adaptec 5405Z controller
• Drives: Seagate Cheetah

When trying to boot into the stock kernel it panics on the device-mapper initialization leading me to believe it was an issue with the RAID driver. It appears the stock kernel has the latest aacraid drivers although I've attempted to re-compile the kernel with the latest from adaptec.com with the same issue.

Here is the stack trace I'm receiving. Any information as to how to further debug this would be greatly appreciated. Thanks!

https://s3.amazonaws.com/uploads.hipchat.com/9460/23303/80l8xsnfl7fa7e9/panic2.png

I really want to run MCollective on my fleet of servers however there are a large number of untrusted users on each machine which makes using MCollective not ideal in my eyes. I'm aware that there is some things you can do to take precaution but I'm not familiar enough with ActiveMQ / want something that's a bit more mindful of similar environments to mine outside the box.

I'm looking for a fact collection like tool essentially.

(Tagging under puppet / server since no mcollective tag and I don't have enough reputation to create a new one)

http://mathias-kettner.de/check_mk.html

I've been testing it out on a couple of development machines and it seems pretty nifty. I cannot however find much information on deployments of it. Does anyone run this actively? Did anyone rule this out as an option for some reason?

I'm having issues with the first mount of eCryptFS. Currently when I attempt to mount it

mount -t ecryptfs ~/.Private ~/Private

However it's erroring. The error in dmesg is the following:

Error parsing options; rc = [-22]

This is a fresh installation of Arch and I've yet to have this issue with other distributions. I'm not quite sure what the issue is however.

Thanks!

Currently we have one beefy MySQL database that runs a couple of high traffic Django based websites as well as some e-commerce websites of decent size. As a result we have a fair amount of large databases using both InnoDB and MyISAM tables.

Unfortunately we've recently hit a wall due to the amount of traffic so I've setup another master server to help alleviate reads / backups.

Now at the moment I simply use mysqldump with a few arguments and it's proven to be fine.. until now. Obviously mysqldump is a slow quick method however I believe we've outgrown its use. I now need a good alternative and have been looking into utilizing Maatkits mk-parallel-dump utility or an LVM snapshot solution.

Succinct short version:

• I have a fairly large MySQL databases I need to backup
• Current method using mysqldump is inefficient and slow (causing issues)
• Looking into something such as mk-parallel-dump or LVM snapshots

Any recommendations or ideas would be appreciated - since I have to re-do how we're doing things I rather have it done properly / most efficient :).