s g's questions

I'm using minikube version 0.28.0 and files don't persist after a restart.

I ran into the problem after noticing my database data disappears after a restart (even though the pods, pv's, and pvc's say they've been up for days). I used minikube ssh to create some files and noticed those disappear, too.

FWIW - I ran minikube addons list to confirm that default-storageclass is enabled

Should I be using minikube --mount to mount my data to my dev machine's disk? How can I persist my pods' data?

I'm looking to do a blue-green(ish) deployment on a single server (single server due to various factors like having a very simple static site, upcoming infra overhaul, among others). Would the following design be adequate for having minimal/no downtime on an apache server?

/var/www/vhosts/site -> ./site-blue
/var/www/vhosts/site-blue/
/var/www/vhosts/site-green/
/var/www/vhosts/swap.sh

For deployment, we would update the inactive site slot then swap the symlink to point to the other color, then restart the server. I'm pretty sure the symlink will work fine (I see a +FollowSymLinks option for the <Directory> tag in the apache conf file) little uncertain about the swap+restart step.

I'm wondering what the underlying architecture looks like for Azure Web Apps for Linux with Node.js. I can't seem to find any information on this in the documentation

I'm curious about things like whether Node sits behind Nginx or Apache, is there any caching layer, is Node.js cluster being used, etc.

What does the architecture look like?

Right now we're hosting our website on an Azure web app. We've created a new repo for a complete overhaul of the site (as opposed to a new branch or something within the existing repo).

The switchover plan was to have two webapps - the exist one and a new one which hosts the new site - and use Azure Traffic Manager to do switch traffic from the old to new site. We tried this and got a "multiple domains point to region" error. We could easily work around this (new app in different datacenter, then clone of new app in original datacenter) but it seems like there should be a better way.

What is the best practice for achieving our goal of switching traffic from one app to another within the same datacenter?

Say I'm debugging an Ansible playbook and want to quit after a given task (and not run through all of the following tasks). Is there any one-line magic command available, or do I have to manually create an exit/assert task?

From the ansible-playbook manual, I see that there is a --start-at-task=START_AT flag, but I don't see anything like an 'end-at' counterpart.

I'm using Ubuntu Server 11.10 x86, and after I make my normal selection (Ubuntu, with Linux 3.0.0-12-generic-pae) from the GRUB menu, the screen just goes blank - there's not even a cursor showing up.

Basically I (suspect that I) had a bad NFS mount tying up the boot process so I fixed that and now it seems like the boot is actually worse off. Here's the problem I had before: Is a bad NFS mount preventing a clean boot?

Not sure if this is useful, but when I hit the 'e' key in the GRUB menu, this is the configuration that shows up:

setparams 'Ubuntu, with Linux 3.0.0-12-generic-pae'

recordfail
set gfxpayload=$linux_gfx_mode
insmod gzio
insmod part_msdos
insmod ext2
set root='(hd0, msdos2)'
search --no-floppy --fs-uuid --set=root <long hex value>
linux /vmlinuz-3.0.0-12-generic-pae root=UUID=<another long hex value> ro
initrd /initrd.img-3.0.0-12-generic-pae

Recovery mode works just fine, networking and all. In fact, if select recovery mode it actually brings up the following menu:

Recover Menu (limited read-only menu)
  resume - Resume normal boot
  fsck - Check all file systems (will exit read-only mode)
  remount - Remount / read/write and mount all other file systems
  root - Drop to root shell prompt

and if I select resume I get the following output:

<timestamp> tpm_tis 00:00: tpm_transmit: tpm_send: error 4294967234
initctl:Event failed

Ubuntu 11.10 <server> tty1
<server> login:

Everything works fine after that.

Other things I've noticed: 1) when booting normally, the Num Lock and Scroll Lock lights on they keyboard show (after pressing the corresponding keys), but the Caps Lock doesn't 2) Ctrl+Alt+Fx to switch TTY's doesn't work 3) this link doesn't help at all: https://wiki.ubuntu.com/X/Troubleshooting/BlankScreen

So, back to the original question: why is my machine not showing anything when booting (in non-recovery mode)?

I'm on Ubuntu server (I think it's 12.04, but not positive) and I basically ran into this problem here Unmount a nfs mount where the nfs server has disappeared

The umount command didn't work, so I tried to just use the ol' reboot. Now the machine gets an error when booting:

FS-Cache:netfs 'nfs' registered for caching
<two minutes later>
init: udevtrigger post-stop process (345) terminated with status 1

I tried booting in recovery mode so I could just comment out the bad NFS mount in my /etc/fstab file, but I wasn't able to write the changes.

I basically just want to get the machine booting properly again so that I can erase the bad mount point, either order is fine.

What are my options here?

I remember messing around with crontab and setting up email capabilities on a server many months back, and now all of a sudden I'm getting the following email:

EMAIL HEADER:

from:    root <[email protected]>
to:  root
date:    Thu, Dec 5, 2013 at 6:48 AM
subject:     Cron <root@server-ip> test -x /usr/sbin/anacron || ( cd / && run-parts --report /etc/cron.daily )
mailed-by:   gmail.com

EMAIL BODY:

/etc/cron.daily/mlocate:
/usr/bin/updatedb.mlocate: `/var/lib/mlocate/mlocate.db' is locked (probably by an earlier updatedb)
run-parts: /etc/cron.daily/mlocate exited with return code 1

This is on Ubuntu Server 12.04.1 LTS. I don't even know which email server I'm using or how to find out. My understanding of the /etc/cron.daily directory must be off, because when I do crontab -l (for both root and my regular user) it says there is no crontab.

Maybe it's an issue with updatedb and not email or cron?

EDIT: I've killed off the /bin/bash /etc/cron.daily/mlocate and /usr/bin/updatedb.mlocate processes and removed the lock, now I'll just have to wait to see if I get the error again.

When running any sudo command from Jenkins I get the following error:

sudo: no tty present and no askpass program specified

I understand that I can solve this by adding a NOPASSWD entry to my /etc/sudoers file which will allow user jenkins to run commands without needing a password. I can add an entry like this:

%jenkins ALL=(ALL)NOPASSWD:/home/vts_share/test/sudotest.sh

...but this leads to the following issue: how to avoid specifying full path in sudoers file?

I can add an entry like this:

%jenkins ALL=NOPASSWD: ALL

...but this allows user jenkins to avoid the password prompt for all commands, which seems a bit unsafe. I'm just curious what my options are here, and if there are any best practices I should consider.

I have a VMDK (with several partitions) file which I need to modify. It is not attached to any particular virtual machine.

I've tried using vmware-mount (from the VDDK toolset) in on Ubuntu server 12.04 which allows me to mount a particular partition from the VMDK to my local machine. I can successfully view the files and, after changing some permissions, I can write changes to the files.

The problem is that they don't persist after I unmount the vmdk (then remount it to check for the changes). Do I need to do anything before unmounting? I'm currently using vmware-mount -d to unmount after making my changes.

I haven't tried to convert the VMDK to a different format, make changes, then convert back. I'm skeptical about going that route.

I am trying to add a NOPASSWD entry for sudotest.sh (or any script/binary that requires sudo) in my /etc/sudoers file (on Ubuntu 12.04 LTS server), but in order to make it work, I must specify the full path. The following entry works just fine:

%jenkins ALL=(ALL)NOPASSWD:/home/vts_share/test/sudotest.sh

The problem is that the script might move to a different directory. This seems like a great chance to use the * wildcard in the path (i.e. /*/sudotest.sh) so that my script could be in any directory but the manual states that wildcards will not match the / character when used in a path. I've confirmed that it doesn't work.

I know that I can use the word ALL in place of my script, but this means there is no password prompt for any commands which seems unsafe.

How do I solve this?