Google Cloud Platform wants to hear from you

[Katie Ball (Google Cloud Support)]

Hi,

My name is Katie, and I am on the Google Cloud Platform technical support team.

This message is to Google Cloud Platform community members, especially if you are newer to GCP. I would like to know what our team can do to help you have a better and more enjoyable experience during the first days on GCP.

Did you need technical support?  If so, I’d like to hear all about it.

I’d also like to know:

• What did you find most difficult about the first-time user experience?

• Where did you get stuck?

Please reply to the group with your answers or any ideas you have on how the technical support team can help new customers get familiar with GCP.

And as a thank you for the great ideas, we will be giving away support coupons worth $450 (equivalent to 3 months of silver support) to 5 lucky community members who post a response. Please make sure to reply before April 8th.

Thanks for your insights, and cloud on!

Katie

[Italo Maia]

Please, make it easier and clearer the support hiring mechanism.
Temporary vm's used in click-to-deploy should not be counted during the free trial. Had quite a few issues with it and took me a while to realize the problem.
It would be nice to have more recipes and tips for GCP. Had a few issues with uwsgi and fabric, two common tools.

[Katie Ball (Google Cloud Support)]

Thanks for the feedback, Italo. This is just the type of information we are looking for! 

Could you clarify what you mean by 'support hiring mechanism'? 

Are you still experiencing any issues or do you have any outstanding questions?

To our other community members, I'd love to hear your feedback about your GCP experience. 

[Leif Pedersen]

Hi Katie, thanks for asking. I have suggestions:

1. Currently afaict, the only way to gain additional publicly accessible IP addresses on an instance is with Network Load Balancing, but this costs $18-19 per month. Network Load Balancing can do a lot more, but when I just need to add an IP that's outrageously expensive. If this is not the case, please let me know!

2. Interactive console. Sometimes we screw up, and sometimes the OS screws up, and right now it's time-consuming to fix. The read-only virtual serial console is only somewhat helpful for gathering information. When a machine crashes, I'd like to be able to break to the kernel debugger so I can get a back-trace, process list, and trigger a kernel core dump for OS developers. Also, if one is locked out from something like network misconfiguration, recovery currently requires killing the machine, mounting it on another VM, fsck (perhaps), fixing the little thing that caused the hassle, and putting it back on the original VM. 90% of this time could be saved with a read-write serial console. I don't actually care if it's serial or how to access it; serial+SSH is the obvious answer, but VGA+VNC would be fine also and would help Windows users more.

3. Add top-tier FreeBSD support. FreeBSD has several features that are superior to Linux. (To avoid the inevitable troll, yes, I acknowledge the reverse also; different jobs, different tools.) It's difficult to bootstrap the first instance without another FreeBSD instance available: one has to jump through time-consuming hoops[1]. Current versions of FreeBSD have improved update and package management tools immensely, so if this was a reason not to support it a few years ago, it should be much more automateable now. I would suggest offering two images, one that uses ZFS for the root filesystem, and one that uses UFS.

[1] FreeBSD Support discussion

- Leif

[Katie Ball (Google Cloud Support)]

Hi Leif,

Thanks so much for raising those points. All three are great points, and it's feedback the rest of the GCP community likely shares.

I have a few follow-ups for you:

For #1, you're right. GCP doesn't currently support multiple publicly available IP addresses. I'm wondering if your interested in running multiple virtual web sites off a single machine? Like with Apache virtual hosts: http://httpd.apache.org/docs/2.2/vhosts/examples.html 

For #3, we've been working with FreeBSD community and are eager to continue to do so. I'd encourage you to sent an email to the appropriate FreeBSD list (I think this one is relevant: freebsd-cloud@freebsd.org) to raise awareness and indicate the need for this request on the FreeBSD side too (as you did here for GCP).

To all our other GCP community members that still have feedback, please let us hear it. We're extending the deadline for our support coupons give away for 1 more week, so it's not too late.

[Leif Pedersen]

On Wed, Apr 8, 2015 at 8:14 PM, Katie Ball (Google Cloud Support) <kmrich...@google.com> wrote:

Hi Leif,

Thanks so much for raising those points.

You bet.

All three are great points, and it's feedback the rest of the GCP community likely shares.

I have a few follow-ups for you:

For #1, you're right. GCP doesn't currently support multiple publicly available IP addresses. I'm wondering if your interested in running multiple virtual web sites off a single machine? Like with Apache virtual hosts: http://httpd.apache.org/docs/2.2/vhosts/examples.html 

That's one use-case, yes. The problem is, the virtual hosting that HTTP supports is (partly) incompatible with SSL, because the server must present the certificate before the client specifies the name of the web site it's looking for. It works well for sites under the same wildcard certificate, but not for hosting multiple domain names. While there are adjustments to the protocol with growing web browser support, last I checked it wasn't compatible with a satisfactory number of browsers. I admit I haven't investigated in a few years.

In my case, I want to run two web servers under different security roles. The web servers need access to different MySQL accounts, which are separated because I host a few web sites for people who don't know each other. To prevent users from seeing each other's passwords by simply looking at each other's document directories, I needed to run separate daemons. I could put a front-end proxy on the public IP and hand off the requests to multiple back-end servers on the same machine, but that's a lot of extra complexity, which means more maintenance and more points of failure and more complex monitoring.

In any case, HTTP isn't the only protocol of interest. Anonymous FTP servers may want to offer different file hierarchies; one may want to host multiple IRC servers on one machine, and VPN servers may want to offer different access, etc.

For #3, we've been working with FreeBSD community and are eager to continue to do so. I'd encourage you to sent an email to the appropriate FreeBSD list (I think this one is relevant: freebsd-cloud@freebsd.org) to raise awareness and indicate the need for this request on the FreeBSD side too (as you did here for GCP).

Indeed, I've been tracking that progress and I'm glad to see it.

Out of curiosity, is there anything in particular that's stalled by the FreeBSD group's efforts? Indeed I use it on GCE personally, and my experience is that it runs well and isn't ungodly slow or anything as a virtualization guest, although I admit I haven't benchmarked it against Linux. They've done some good work specifically on general virtualization. The only thing really stopping it from top-tier support on GCE appears to be Google's special sauce: A public image available for installation, integration of GCE tools, etc.

To all our other GCP community members that still have feedback, please let us hear it. We're extending the deadline for our support coupons give away for 1 more week, so it's not too late.

Oh no! I hoped I'd have a 50-50 chance of that prize. ;)

- Leif

On Tuesday, April 7, 2015 at 9:30:35 AM UTC-7, Leif Pedersen wrote:

Hi Katie, thanks for asking. I have suggestions:

1. Currently afaict, the only way to gain additional publicly accessible IP addresses on an instance is with Network Load Balancing, but this costs $18-19 per month. Network Load Balancing can do a lot more, but when I just need to add an IP that's outrageously expensive. If this is not the case, please let me know!

2. Interactive console. Sometimes we screw up, and sometimes the OS screws up, and right now it's time-consuming to fix. The read-only virtual serial console is only somewhat helpful for gathering information. When a machine crashes, I'd like to be able to break to the kernel debugger so I can get a back-trace, process list, and trigger a kernel core dump for OS developers. Also, if one is locked out from something like network misconfiguration, recovery currently requires killing the machine, mounting it on another VM, fsck (perhaps), fixing the little thing that caused the hassle, and putting it back on the original VM. 90% of this time could be saved with a read-write serial console. I don't actually care if it's serial or how to access it; serial+SSH is the obvious answer, but VGA+VNC would be fine also and would help Windows users more.

3. Add top-tier FreeBSD support. FreeBSD has several features that are superior to Linux. (To avoid the inevitable troll, yes, I acknowledge the reverse also; different jobs, different tools.) It's difficult to bootstrap the first instance without another FreeBSD instance available: one has to jump through time-consuming hoops[1]. Current versions of FreeBSD have improved update and package management tools immensely, so if this was a reason not to support it a few years ago, it should be much more automateable now. I would suggest offering two images, one that uses ZFS for the root filesystem, and one that uses UFS.

[1] FreeBSD Support discussion

- Leif

--
© 2014 Google Inc. 1600 Amphitheatre Parkway, Mountain View, CA 94043
 
Email preferences: You received this email because you signed up for the Google Compute Engine Discussion Google Group (gce-dis...@googlegroups.com) to participate in discussions with other members of the Google Compute Engine community and the Google Compute Engine Team.
---
You received this message because you are subscribed to the Google Groups "gce-discussion" group.
To unsubscribe from this group and stop receiving emails from it, send an email to gce-discussio...@googlegroups.com.
To post to this group, send email to gce-dis...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/gce-discussion/a97b7846-3ed1-4666-9a19-4f8c5b3f264c%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--

As implied by email protocols, the information in this message is not confidential.  Any middle-man or recipient may inspect, modify, copy, forward, reply to, delete, or filter email for any purpose unless said parties are otherwise obligated.  As the sender, I acknowledge that I have a lower expectation of the control and privacy of this message than I would a post-card.  Further, nothing in this message is legally binding without cryptographic evidence of its integrity.

http://bilbo.hobbiton.org/wiki/Eat_My_Sig

[Bryan Hiestand]

• What did you find most difficult about the first-time user experience?

Initial gcloud authentication with the SDK on OS X apparently requires port 8085.  I already have a local service using this port, so initial authorization/authentication to my google account failed.  This should be clearly documented on the quickstart guide or prompted by the installer.

To add a second "most difficult", many search results along the lines of "google compute engine move instance" return docs referencing the deprecated gcutil.  Hopefully gcutil links will continue to drop in the rankings.

Now that I am comfortable with them, I have to say I am very impressed with the interface and provided utilities.

My biggest concern right now is outages.  Specifically, I was affected by Incident 15050 for ~6 hours today, and experienced similar packet loss last weekend.  I realize I am still in my free trial, but customers should be able to report outages.  My users were complaining about packet loss 3 hours before the Cloud team even acknowledged the issue.  Paying for the ability to report outages may be a dealbreaker for me.

[Kapil Thangavelu]

On Thu, Apr 2, 2015 at 12:06 PM, Katie Ball (Google Cloud Support) <kmrich...@google.com> wrote:

Hi,

My name is Katie, and I am on the Google Cloud Platform technical support team.

This message is to Google Cloud Platform community members, especially if you are newer to GCP. I would like to know what our team can do to help you have a better and more enjoyable experience during the first days on GCP.

Did you need technical support?  If so, I’d like to hear all about it.

I’d also like to know:

• What did you find most difficult about the first-time user experience?

That the gce console is broken with browsers that are logged into multiple google accounts.

cheers,

Kapil

[Sergio Garcia Murillo]

+1, just created a ticket to GCE support, it didn't failed for me before...

BR
Sergio

[Katie Ball (Google Cloud Support)]

Thank you all for the feedback. These are exactly the ideas that will improve GCP.

I'm sorry to hear that you were affected by Incident 15050. You raise a very good point regarding the ability to report incidents to our team. I've submitted a feature request to the appropriate team on your behalf. 

Just as a reminder, if you believe this incident impacted GCE's 99.95% monthly uptime percentage SLA (as defined here), please file a ticket using the Google Developers Console Billing Support Requests form to notify us of SLA financial credit eligibility.

Thanks again for the great ideas.  We really appreciate your suggestions!

[Joe Pierce]

Hello Katie,

I used the Bitnami launch pad to launch a WordPress Multisite on the Google cloud. The installation was effortless and I quickly had my multisite up and running. Everything is working and I was able to figure out most things but I have a couple important questions that I was unable to answer.

How, exactly, can I reliably backup my database using the snapshot feature? I made a test installation and took a snapshot. I then made some edits to the contents of the site so I could see if my attempt to restore the site to the snapshot was successful. Never got that far...  I attached the snapshot to a new persistent disk but I could not find where or how to attach this new disk as a boot drive to restore the site. Backups are not an optional capability, I must find a solution in order to use GCE in my projects.

Please don't send me a bunch of links that only touch upon this issue, as I have most likely already read them. Backing up a project is a very basic issue that every GCE user must deal with. I realize that the Google model is to be all things to all people, but that approach leaves me in a blur of too much data and conflicting advice. Please, oh please, show me a simple path! What I need (please) is a very straight forward, numbered list of specific recommended backup procedures and code lines that directly apply to a WordPress MultiSite installation like mine.

Also, I read in http://googlecloudplatform.blogspot.com/2013/10/persistent-disk-backups-using-snapshots.html that states:

 

"First, we strongly recommend using mounted volumes for application data (not boot disk). Among other things, it’s easier and safer to take snapshots in a running instance." 

How would I accomplish this? Is this something that applies to my installation?

Background: My installation uses MySQL Server version: 5.5.42 Linux 3.16.0-0.bpo.4-amd64 #1 SMP Debian 3.16.7-ckt9-2~bpo70+1 (2015-04-21) x86_64 WordPress is version 4.2. My work station is OSX 10.9.5. I am able to connect a SSH using Terminal to the SFTP directories and the Bitnami control panel provides a SSH window where I am able to access the project's back end at the bitnami user level.

Thank you very much, Katie. I look forward to your response. btw~ I would LOVE to get one of your support coupons!

Cheers!

[Jesse Scherer (Google Cloud Support)]

Hi Joe,

I'll try to answer each of your questions individually. First a disclaimer: none of this advice is specific to WP Multisite but instead the task you're trying to accomplish: full backups.

When you tried to restore from backup before, it sounds like you tried to attach a new boot disk to an existing instance. If so, my advice is to create a new instance and attach the new disk which you describe having created as that new instance's boot disk. In other words, don't worry so much about re-using the same instance.

I hope I'm not frustrating you by sending you another link, but this discussion on Bitnami's forum is very directly related to this task: https://community.bitnami.com/t/increasing-disk-size-of-a-live-instance-hosted-through-google-cloud/30629. The only difference is that you are not (necessarily) getting a bigger disk, but the logic is the same: 

- take a backup,

- create a boot disk from it, 

- create and start an instance with that disk

You second question is about having a separate volume for application data. As far as I know, this is not how Bitnami sets up instances. However at the bottom of the same thread, use "vipul" describes how they accomplished this.

- create a new disk

- attach it to your instance

- copy data to new disk

- replace the data directory with a symlink to the new volume

In the first case where everything is on a root disk, you must shut down your instance to do the backup in order to prevent filesystem corruption. In the second case, you would still need to freeze the disk being backed up, which implies some downtime for your site. 

Again, without being aware of any WP Multisite specific gotchas, I hope this helps.

[Jesse Scherer]

Don't worry about picking an image. The backup you created should contain everything you need to boot up the OS, since Bitnami is putting everything: OS, configuration, Wordpress software and your data into the same disk.

On Tue, Apr 28, 2015 at 4:17 PM, Joe Pierce <j...@joe-pierce.com> wrote:

Hello Jesse,

Thanks for your quick response. I appreciate it! 

I think the fog is starting to lift.... I should just create a new instance of a Debian server using the snapshot to provide all the files unique to my server, then I connect that new server to my domain name and blast the original. How do I connect the new server to my domain name?

I looked at the available disk images in the admin section. It is a Chinese menu to my eyes. Which one should I choose? Which one would you choose?

Cheers!

--

© 2014 Google Inc. 1600 Amphitheatre Parkway, Mountain View, CA 94043
 
Email preferences: You received this email because you signed up for the Google Compute Engine Discussion Google Group (gce-dis...@googlegroups.com) to participate in discussions with other members of the Google Compute Engine community and the Google Compute Engine Team.
---

You received this message because you are subscribed to a topic in the Google Groups "gce-discussion" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/gce-discussion/q2Lp4_4mf5Y/unsubscribe.
To unsubscribe from this group and all its topics, send an email to gce-discussio...@googlegroups.com.

To post to this group, send email to gce-dis...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/gce-discussion/ff0bf5a5-267b-44c9-8021-5c163381584e%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--

Jesse J. Scherer
Technical Program Manager, Google Cloud Support
jsch...@google.com
+1 (404) 803-0804

[se...@bangmediagroup.com]

Changing machine type could be easier. I would like to be able to change the machine type from one to another with ease. Currently I can use gcloud delete and add it again. I cannot easily do this from the web interface, I need to delete and wait until the UI stops telling me the a system with that name already exists.

Query caching on CloudSQL is disabled which is rather important for many of our use cases. This was a surprise and resulted in having to use MySQL in GCE instances. 

Finding out if this was disabled on purpose / if we can have it enabled was problematic. 

Happy with the performance of the GCE instances and Big Query.

[Sergio Garcia Murillo]

Hi Katie,

Not sure if it is exactly the kind of feedback you were asking for, but anyway.. :)

One of the critical missing pieces of GCE IMHO is a good way to have a shared storage between instances. AFAIK there are only three ways of doing it currently:

• Mount a disk RO shared between instances, this is obviously a no go
• Use google cloud storage for storing files, which  works, but it is quite cumbersome, as we need to add extra processes to send and retrieve files from it and do custom integrations, which is not always possible.
  
• Building your own solution, like NFS share, which adds extra operational work and requires extensive knowledge to do it right (HA, backups, raids, etc)

It would be awesome if GCP provided that functionality natively.

Best regards
Sergio

--
© 2014 Google Inc. 1600 Amphitheatre Parkway, Mountain View, CA 94043
 
Email preferences: You received this email because you signed up for the Google Compute Engine Discussion Google Group (gce-dis...@googlegroups.com) to participate in discussions with other members of the Google Compute Engine community and the Google Compute Engine Team.
---

You received this message because you are subscribed to the Google Groups "gce-discussion" group.

To unsubscribe from this group and stop receiving emails from it, send an email to gce-discussio...@googlegroups.com.

To post to this group, send email to gce-dis...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/gce-discussion/efb74e95-12c6-4e51-8273-df8656e31dec%40googlegroups.com.

[Alexander Lücking]

Hi Katie,

we evaluated GCP the last week and unfortunately had a very bad experience.

We installed two VM's with Elastic Search and one App Engine for image manipulation and routed some live traffic to the Elastic Search instances. We used the free trial option.

One day later the VM's got shut down by Google for possible "Denial of service attacks"

"Google has found activity on ... that appears to be committing denial of service attacks."

We have no idea how Google came to this conclusion.

Now, 6 Days later (the appeal form states: "You will receive a response within 2 business days."), we still haven't got any response from support and GCP keeps being blocked.

How can this be improved:

• Warn customers beforehand when their entire farm is getting shut down.
• Respond to filled out appeal forms in the stated time or faster

Pixum App

Pixum Fotobuch | Pixum Wandbild Simulator | Pixum Service | Pixum Instacard

Jetzt herunterladen: www.pixum.de/apps/

artbox44.com

Kunst & Design für Ihre 4 Wände.

Jetzt inspirieren lassen und online bestellen: www.artbox44.com

Pixum - Immer die schönsten Bilder

---------------------------------------------------------------------------------------------------------------------
Pixum ist eine geschützte Marke der
Diginet GmbH & Co. KG - Industriestr.161 - 50999 Köln

Fon: +49 (2236) 886-0 - Fax: +49 (2236) 88 66 99

Sitz Köln, HRA 25531, Umsatzsteuer-ID: DE-209867661
Komplementärin: Diginet Management GmbH, Sitz Köln, HRB 69766  
Geschäftsführer: Daniel Attallah, Oliver Thomsen
---------------------------------------------------------------------------------------------------------------------
www.pixum.de  .at .be .ch .com .uk .dk .es .eu .fr .nl .no .pl ...

[Jesse Scherer]

Joe, at a glance those steps look right. Also moving to a bigger instance should be no problem. Do exactly the steps you described above, but when you create a new instance, make one that is bigger than the instance you created the snapshot on.

I do the opposite of this often: I will set up a new server on a big instance and then downsize based on actual observed usage.

On Thu, Apr 30, 2015 at 11:05 AM, Joe Pierce <j...@joe-pierce.com> wrote:

G'morning, Jesse,

The google site restore from snapshot goes like this:

• Login to google console. Click on VM instances.
  
• Select the instance you want to snapshot and click stop.
  
• Click on snapshot and create a snapshot.
  
• Go back to VM instances, click on the instance and click start from the instance settings page. Return to the VM instance list page.
  
• Click new instance. name and provision the instance and click change next to the disk description.
  
• Click on the snapshot heading and select your snapshot and save.
  
• Delete the original disk and click on the new disk.
  
• Change the external IP from the default ephemeral address to your static ip and save.
  

That's it!

If you see anything in this process that I should change, please advise.

The process was really easy using the Developers console. All of the documentation I found references command line code. Is there any documentation that focuses on the console? In particular, is it possible to put a snapshot on a bigger instance using only the Developer console?

Thanks again, Jesse.

On Wed, Apr 29, 2015 at 1:22 PM, Joe Pierce <j...@joe-pierce.com> wrote:

Hello again, Jesse.

I am currently swamped with other stuff just now, so it may take a few days before I can work this through and respond. I will apply what you sent and report back in detail. I think this conversation will lead to a useful process that many GCP users will reference and appreciate.

Thank you very much for your help and attention, and hangin' with me through this issue....

Cheers!

Joe

On Thursday, April 2, 2015 at 10:06:56 AM UTC-6, Katie Ball (Google Cloud Support) wrote:

Hi,

My name is Katie, and I am on the Google Cloud Platform technical support team.

This message is to Google Cloud Platform community members, especially if you are newer to GCP. I would like to know what our team can do to help you have a better and more enjoyable experience during the first days on GCP.

Did you need technical support?  If so, I’d like to hear all about it.

I’d also like to know:

• What did you find most difficult about the first-time user experience?

• Where did you get stuck?

Please reply to the group with your answers or any ideas you have on how the technical support team can help new customers get familiar with GCP.

And as a thank you for the great ideas, we will be giving away support coupons worth $450 (equivalent to 3 months of silver support) to 5 lucky community members who post a response. Please make sure to reply before April 8th.

Thanks for your insights, and cloud on!

Katie

--

© 2014 Google Inc. 1600 Amphitheatre Parkway, Mountain View, CA 94043
 
Email preferences: You received this email because you signed up for the Google Compute Engine Discussion Google Group (gce-dis...@googlegroups.com) to participate in discussions with other members of the Google Compute Engine community and the Google Compute Engine Team.
---

You received this message because you are subscribed to a topic in the Google Groups "gce-discussion" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/gce-discussion/q2Lp4_4mf5Y/unsubscribe.
To unsubscribe from this group and all its topics, send an email to gce-discussio...@googlegroups.com.

To post to this group, send email to gce-dis...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/gce-discussion/ff0bf5a5-267b-44c9-8021-5c163381584e%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--

Jesse J. Scherer
Technical Program Manager, Google Cloud Support
jsch...@google.com
+1 (404) 803-0804

[petit curieux]

Hi ,

don't know if its usefull feedback , but i think the gui firewall rules could be better , actually 

i didn't find it very intuitive . better gui could be easier to manage rules ..

[Joe Pierce]

There are 30 sites on my WP multisite project that, at this point in development, have exactly the same data set, except each site is using different a theme. I put all sites in a (chrome) bookmark folder then, as an experiment,  I opened them all at the same time to see how the server responded. It took about 3 minutes to open all of them! Below are screen shots from my google developer console that clearly document the slam on the server.

The project is on a google cloud platform g1-small instance (1 vCPU, 1.7 GB memory) using a Bitnami image. Storage is on a 222 GB Standard persistent disk. The entire multisite server directory is currently only 675 MB.

Once the project goes operational I hope to have hundreds of sites generating lots of traffic (and funding to expand capacity). I want to be able to handle my server load efficiently, but I don't want to pay any more than necessary while I am getting started. In the meantime, the server must be as fast as possible in order to help me start selling sites. (No one wants to buy into a slow site.)

My questions is, which instance would best fit for this project? 

Thank you.