Has anyone ever done an upgrade to the same version?

By that, I mean, installed, say XS7 over an existing XS7 install?

I am thinking of doing that to test my "clone to a bigger USB" theory.

Another question, this one more for general knowledge.

If I were to clone the current boot USB, and then try this upgrade and it bombs, can I just go back to the cloned original setup?

This is the kind of stuff that I don't think you can do, but then @scottalanmiller says ... "you are thinking windows ... Linux is just .... EASY"

@travisdh1 said

That's within reason for a 100Mb network connection. The best I've seen out of a 100Mb network is 9MB/sec, that's on a network with zero other traffic.

Yeah, I am really just chatting at this point.

I'll set up two VMs on separate segments and test straight throughput and then the export/copy ... everything else.

@travisdh1 said in XenServer 7 has launched!:

@BRRABill 1Gb networks are much nicer, and not any more expensive anymore. For future consideration.

Yeah, it's a long story.

The servers are all connected via 1GB, though.

@travisdh1 said

Assuming we both make it to Mangocon, maybe we can trade war stories

Sounds good!

I mean, 100Mbps is fine for 99% of what we do. It's just these VM transfers are a bear.

But, the new me with my fancy ML knowledge now just sets up a Linux instance and does the import/export on the 1GB segment. Viva la open source!

Of course THAT was also slow, which has me wondering... But as I said, nothing firm tested yet so we shall see.

Ah, I see I do. Thank you Google.

It was 40 Mbps. Or around 5 MBps.

@olivier said

@BRRABill Replication uses export from one XS and pipe it to the import of another one. So in any case, it's import/export.

OK, when I get my second XS7 stood up, I will run a test from XO and let you know.

Still, seems very strange that everything is copying perfectly EXCEPT exporting from XS, especially given all the other people with issues.

I wonder why yours is working. "What's your secret????" I ask, wanting to hurry these exports the heck up!

If anyone else has XS7 set up already, if you get some time, export a small VM and see what speeds you get.

@momurda said

I was thinking perhaps you are hitting a bottleneck here if your settings force this go through a router that doesnt do Gb.
I hit about 400mbps (not constantly, but consistently) doing exports this way, during middle of the day. I could test when nobody is around later and hit a bit higher througput.

I finally got iPerf to work on the XS host.

Here are the results. Pretty sure it's not the connection.

[ 3] local 10.0.4.30 port 55737 connected with 10.0.4.40 port 5001
[ ID] Interval Transfer Bandwidth
[ 3] 0.0-10.0 sec 1.09 GBytes 938 Mbits/sec

@aaronstuder said in You know you have been...:

@BRRABill ip addr

Hmmm, I think I like ifconfig better!

@momurda said

Definitely not

I got to thinking ... @olivier said that most of the stuff flows through XO. Like the copy is basically an export -> XO -> import ... I wonder if my XO instance isn't beefy enough.

We set it up minimal.

I'm going to test that after the July 4th holiday weekend.

The other key tip is to move the logging off the USB.

Did we ever decide if the swap file was a danger as well? (A danger to prematurely burning out the USB cells.)

Boy once I reconfigured the right click in PUTTY it was so much easier to use.

@travisdh1 said i

Micro Center has 64GB for $13. Best warranty you'll find is the real reason I like those tho, if it ever goes bad they'll replace it.

I'm a little OCD so I've been buying the little teeny ones that go flush on the server.

But your link looks like a great option.

As a point of reference...

I did a live migration tonight. So, basically the same sort of thing ... copying a VM from one XS to another.

It clipped along at about 65MBps. Took a little over 30 minutes to migrate about 110GB of data.

So, I don't think it is the hardware, or the network, or anything like that. It must truly be some bug in the exporting...

@scottalanmiller said

That doesn't seem to highlight where it is failing, though.

The issue I was having (and the commenter was also having) is that after I made those changes, the files would just overwrite themselves again.

The only way to fix it was to follow the "dirty, dirty trick" he recommends by changing the permissions to read only on those files.

That finally did it for me.

I keep bringing up this issue because as a community we always tell people to install XS and install it on USB. But XS does writing to the boot partition sometimes even when you think it is not.

Has anyone actually LOOKED at their USB boot device. Are you SURE it isn't being written to? Because mine definitely was, and I think that is the standard.

I think it was @DustinB3403 who said to just remap the log directory elsewhere. I guess that would work as well?

And I know you two were discussing the swap partition. Did we ever decide if it really WAS writing to it?

Is there a way to know how much data is being written to a particular partition?

@dafyre said

You need to check the options in whatever app you are connecting with, and make sure that you have the window size set for "fit to window" or whatever the right option is. I use Pyhoca-GUI and don't have to mess with that setting.

Strangely enough, it is working now.

COMPUTERS!

@scottalanmiller said

Linux Mint is a full distro, like CentOS, Fedora, OpenSuse Leap, Ubuntu, etc. It is a distro that focuses on desktop usage rather than server or mixed. But it is a distro. The desktop of Mint is Cinnamon, Mate, LXDE, etc.

How does a distro "focus on desktop usage", exactly?

I think that file has been replaced with /etc/rsyslog.d/xenserver.conf

Which reads...

# Suppress duplicate messages and report "Last line repeated n times"
$RepeatedMsgReduction on

# Don't rate-limit messages - this isn't the right way to go about 
# reducing log size!
$IMUXSockRateLimitInterval 0
$SystemLogRateLimitInterval 0

# Ensure critical and higher level errors are logged synchronously.
*.crit;mail.none;authpriv.none;cron.none		/var/log/crit.log

# Log by facility.
kern.*							-/var/log/kern.log
daemon.*						-/var/log/daemon.log
user.*							-/var/log/user.log

# The authpriv file has restricted access.
authpriv.*						-/var/log/secure

# Log all the mail messages in one place.
mail.*							-/var/log/maillog

# Log cron stuff
cron.*							-/var/log/cron

# Save boot messages also to boot.log
local7.*						/var/log/boot.log

# Xapi rbac audit log echoes to syslog local6
local6.*						-/var/log/audit.log

# Xapi, xenopsd echo to syslog local5
local5.*						-/var/log/xensource.log

# V6d echo to syslog local4
local4.*						-/var/log/v6d.log

# xenstore access to syslog local3
local3.info						-/var/log/xenstored-access.log

# Storage Manager to syslog local2
local2.*						-/var/log/SMlog

# xcp-rrdd-plugins (info and above) to local0
local0.info						-/var/log/xcp-rrdd-plugins.log

# ignore default rules

*.* @10.0.4.26
*.* ~

@DustinB3403

Well, in 6.5, the syntax was as follows...

# Save boot messages also to boot.log
local7.*             @10.0.0.1
# local7.*         /var/log/boot.log

In this new version, it just puts the IP at the bottom.

@scottalanmiller mentioned, as I think you did as well, that I should just redirect /var/log somewhere else.

Being a little fresh in Linux partitioning, I most post a thread for help with that.

@Danp said in Stopping XenServer From Writing To A USB Boot Drive:

Here's one to watch from the Citrix forums: http://discussions.citrix.com/topic/379454-booting-xenserver-off-usb-safe/

P.S. Which one of you guys wrote this?

That was me.

Those guys know the nuts and bolts of XS, though they aren't real responsive to threads all the time.

On ML that would have have 500 posts already and been forked 6 times.