Sabya's questions

I am using Amazon Spot Instances to crawl a lot data. Most of them run until Amazon terminates them once the current price exceeds our max bid.

I need to monitor and mainly archive the logs generated in those spot instances. Those logs are very important for debugging and analytics. We have application logs, system logs such as syslog, secure log. Below are the options I could think of:

• use Chukwa/Flume. Not listing Facebook's scribe here because I think the project is dead. There is a rare possibility to lose few logs with this approach.
• Attach an EBS volume to those spot instance. But then managing those volumes when the spot instances are terminated will be a pain.
• Mount a NFS volume so that we write logs in that volume. The performance is really bad sometimes in this approach.

Also, I think the ability to run Linux commands such as grep, awk on those archived files are also important. What are people using in such situation?

P.S. We are already using splunk but I will not archive logs in splunk.

Possible Duplicate:
SysAdmin & Developer: Responsibilities

Suppose, I have 20 servers:

• We keep data in Linux servers
• Developers often need to Login to the server to debug some issue
• Sometimes they have to access user data and run through the app in production to replicate a problem that was not reproducible in test environment

What are the best practices for this situation?

• I am using Linux (3.2.12-3.2.4).

• I have few "syncer" processes which updates lots and lots SQLite db files in a loop. The database sizes vary from 1 MB to few GB.

• I have a "searcher" process which can query those DBs and return matched results.

I am looking forward to tune some kernel parameters to get maximum IO performance.

GlusterFS still offers the open source software. But since Redhat has re-branded GlusterFS as Redhat Virtual Storage Appliance, I am confused about what will be good for production environment.

Is there any technical difference between Redhat VSA and GlusterFS?

I know that Redhat VSA will come with support but what I am talking about is technical feature differences.

I have created a raid10 array using mdadm. Metadata version is 1.2. I want to convert this to a raid0 array, is it possible?

P.S. I am having terabytes of data, I want to avoid copying the data over.

I want to collect samples of iostat's svctm, await every 5 mins from all of my servers and store them in nagios. I want to get the values for what is happening in every 5 minutes (not since boot time, iostat's first output gives values since boot time). How can I do it in nagios?

EDIT The tps should NOT be calculated #of transactions happened since reboot divided by uptime. What I want is # of transferred happened in last X mins divided X*60.

Our application uses EBS to store data. The storage requirement of the application grows rapidly as users signup. We need to do it online so that users don't experience downtime with the application. So, here is what we do:

• The file system is XFS.
• We are using LVM and RAID0.
• Whenever we fill up 40% of total storage, we add two EBS volumes, create a RAID0 array out of them. Then we add this array to the LVM to extend the file system.

This way, the application can continue to read and write to the file system, while the underlying file system grows.

Yesterday, we experienced corruption in some of the database files that were being written into.

My question is:

• is it safe to use RAID0 with LVM and XFS?
• is it safe to use xfs_grow to increase storage while the files in the file system are being written into?

I am using ngnix to route traffic to proper application servers based on a cookie value. So one user always lands in a particular uptstream server.

Now I have multiple such uptsream servers. I want to send 503 for a upstream server when I am taking it down for maintenance purpose. What is the simplest way to do it?

If the application server is crashed we should get normal "could not connect to backend" error. So, I should get 503 for a upstream only when I am taking it down intentionally.

Our infrastructure is on AWS. I want to get a daily report on how much spent on the previous day. What is the best way to do it?

I'm using nginx, and want to implement SSL session resumption. How should we I test if it is working?

I have enabled these settings:

ssl_session_cache    shared:SSL:10m;
ssl_session_timeout  10m;

The Apache web server in my RHEL shows many requests in R (..reading..) state. I want to find the client IPs that are making Apache wait in reading state. Specifically: I want to find out all the client IPs that are spending too long to send the request.

The server-status module is not enough. Server status does not show the client information when the PID is in reading state.

Our applications are hosted in Rackspace dedicated servers. We are planning to move them to AWS services like EC2, S3, etc. Currently Rackspace takes care of backing up data, monitoring stuff, I don't know what else.

So, my questions is what are the things that we should do ourselves or outsource to other companies like Percona? Can someone list down the things?

I want to use openssl s_client for accessing Gmail POP3S server. I am doing the following:

$ openssl s_client -connect pop.gmail.com:995
+OK Gpop ready for requests from <removed: ip> d11pf35377217wam.36
USER <removed: [email protected]>
+OK send PASS
PASS <removed: password>
+OK Welcome.
LIST
+OK 1 messages (2197 bytes)
1 2197
.
STAT
+OK 1 2197
RETR 1
RENEGOTIATING

The problem is I am never able to execute the RETR command. It always hangs while "RENEGOTIATING".

What am I missing?

I want to use ActiveSync inside a firewall from iPad. What port does ActiveSync communicate on? Which ports do I have to keep open?

What is the process that you guys follow to maintain authentication details like login ids and passwords? There will be definitely some shared passwords. So, the target is to minimize the impact when someone is leaving the company.

By "shared password", I mean, the account, which is shared among multiple people in the company.

The issues that the process should address are: -

1. Affected areas. Quickly find the resources to which the leaving user was having access to.

2. Forgetting password. What happens if a user forgets an authentication details? How does he get it? I think he shouldn't ask a team mate. I mean no-verbal communication.

3. Find dependencies of a resource. Suppose I am changing the password for a mail account, which is getting used by some automated scripts to send mails. Here, the scripts are dependent on the mail account, so changing the password of the mail account means we have to change the password in the script too. So, how do find all the dependencies of a resource?

I'd prefer a process which addresses these issues. But you can also recommend products which are open source and not hosted. I have gone through PassPack, but they don't solve #4.

There is a similar question here. But that does not exactly answer my question.