Posts Tagged ‘directories’

Why SSH Login Is Slow on Linux and How to Fix It

Thursday, December 18th, 2025

openssh_debug-and-fix-slow-connections-to-linux-how-to-original-openssh-logo

Slow SSH logins are one of those problems that don’t look serious at first — until you realize every connection takes 20–30 seconds to respond. The shell eventually appears, but the delay is long enough to break automation, frustrate users, and make admins suspicious of deeper system issues.

This article walks through some common causes of slow SSH logins and how to diagnose them efficiently on Linux servers.

1. DNS Lookups: The Most Common Culprit

By default, sshd performs a reverse DNS lookup on the connecting IP address. If DNS is misconfigured or unreachable, SSH will wait.

How to Test

From the server (measure how many seconds it takes to do ssh to the machine):
 

$ time ssh localhost

If localhost logins are instant but remote logins are slow, suspect DNS.

Check /etc/ssh/sshd_config:

UseDNS yes

Fix

Disable DNS lookups (at least temporary to test):

UseDNS no

Then restart SSH:

# systemctl restart sshd

Note: This does not reduce security in most environments and is safe for the majority of servers.

2. Broken or Slow PAM Modules

PAM (Pluggable Authentication Modules) can introduce delays — especially if modules depend on:

  • LDAP
  • Kerberos
  • Network home directories
  • Smart card services

Debug with Verbose SSH

From the client:

$ ssh -vvv user@remote-server

Look for pauses during:

debug1: Authentications that can continue:

Test PAM Delay

Temporarily disable PAM in /etc/ssh/sshd_config:

UsePAM no

Restart SSH and test again.
If login becomes instant, inspect /etc/pam.d/sshd.

3. Entropy Shortage on Virtual Machines

Older kernels or low-activity VMs can run out of entropy, causing SSH key operations to block.

Check Entropy Level

# cat /proc/sys/kernel/random/entropy_avail

Values below 100 may cause delays.

Fix

Install an entropy daemon (if on Deb based distro):

# apt install haveged

or on CentOS / RHEL / Fedora

# yum install rng-tools

Then start the service:

# systemctl enable –now haveged

4. GSSAPI Authentication Delay

SSH attempts Kerberos authentication even when not used.

Symptom

Delay occurs before password prompt appears.

Fix

Edit /etc/ssh/sshd_config:

GSSAPIAuthentication no

GSSAPICleanupCredentials no

Restart SSH afterward.

5. Slow Home Directory or Shell Initialization

Sometimes SSH is fast, but the shell is slow.

Test with a Minimal Shell

$ ssh user@server /bin/sh

If this is instant, check:

  • .bashrc
  • .profile
  • .bash_logout

Common mistakes:

  • Network calls (curl, wget)
  • Mounted NFS home directories
  • Broken PATH exports
  • Commands waiting on unavailable resources

6. Logging and Timing the Login Process

Enable SSH debug logging in /etc/ssh/sshd_config:

LogLevel DEBUG

Then watch logs:

# journalctl -u sshd -f

or:

# tail -f /var/log/auth.log

This allows you to see exactly where the delay happens.

7. A Systematic Troubleshooting Checklist

  1. Disable DNS lookups (UseDNS no)
  2. Disable GSSAPI
  3. Test PAM
  4. Check entropy
  5. Test minimal shell
  6. Review auth logs

In practice, 90% of slow SSH issues are DNS or PAM related.

Conclusion

But wait there might be much more behind the SSH slowness such as misconfigured LDAP or other infrastructure in the middle.
Slow SSH logins are rarely “just SSH.", and though this guide should help you with some sporadic random server issues, if the issues is present on a complex infra with multiple ssh servers, then  that is almost always a symptom of:

  • Network misconfiguration
  • Over-engineered authentication
  • Broken assumptions about system dependencies

Approaching the problem methodically saves hours of guesswork and restores what SSH is supposed to be, work without glitches.

Delete empty files and directories under directory tree in Linux / UNIX / BSD

Wednesday, October 21st, 2020

delete-empty-directories-and-files-freeup-inodes-by-empty-deleting-directoriers-or-files

Sometimes it happens that you end up on your server with a multiple of empty files. The reason for that could be different for example it could be /tmp is overflown with some session store files on a busy website, or due to some programmers Web executed badly written PHP / Python / Perl / Ruby code bug or lets say Content Management System ( CMS ) based website based on WordPress / Joomla / Drupal / Magento / Shopify etc. due to a broken plugin some specific directory could get filled up with plenty of meaningless empty files, that never gets wiped out if you don't care. This could happen if you offer your users to share files online to a public sharing service as WebFTP and some of the local hacked UNIX user accounts decides to make you look like a fool and run an endless loop to create files in your Hard Drive until your small server HDD filesystem of few terabytes gets filled up with useless empty files and due to full inode count on the filesystem your machine running running services gets disfunctional …

Hence on servers with shared users or simply webservers it is always a good idea to keep an eye on filesystem used nodes count by system are and in case if notices a sudden increase of used FS inodes as part of the investigation process on what caused it to check the amount of empty files on the system attached SCSI / SSD / SAS whatever drive.
 

1. Show a list of free inodes on server


Getting inodes count after logged is done with df command

root@linux-server:~# df -i
Filesystem        Inodes   IUsed     IFree IUse% Mounted on
udev             2041464     516   2040948    1% /dev
tmpfs            2046343    1000   2045343    1% /run
/dev/sdb2       14655488 1794109  12861379   13% /
tmpfs            2046343       4   2046339    1% /dev/shm
tmpfs            2046343       8   2046335    1% /run/lock
tmpfs            2046343      17   2046326    1% /sys/fs/cgroup
/dev/sdc6        6111232  6111232   0   100% /var/www
/dev/sda1       30162944 3734710  26428234   13% /mnt/sda1
/dev/sdd1      122093568 8011342 114082226    7% /backups
tmpfs            2046343      13   2046330    1% /run/user/1000

 

2. Show all empty files and directories count

 

### count empty directories ### root@linux-server:~# find /path/ -empty -type d | wc -l

### count empty files only ### root@linux-server:~# find /path/ -empty -type f | wc -l

 

3. List all empty files in directory or root dir

As you can see on the server in above example the amount of inodes of empty inodes is depleted.
The next step is to anylize what is happening in that web directory and if there is a multitude of empty files taking up all our disk space.
 

root@linux-server:~# find /var/www -type f -empty > /root/empty_files_list.txt


As you can see I'm redirecting output to a file as with the case of many empty files, I'll have to wait for ages and console will get filled up with a data I'll be unable to easily analyze

If the problem is another directory in your case, lets say the root dir.

root@linux-server:~#  DIR='/';
root@linux-server:~# find $DIR -type f -empty > /root/empty_files_list.txt

4. Getting empty directories list


Under some case it might be that the server is overflowed with empty directories. This is also a thing some malicious cracker guy could do to your server if he can't root the server with some exploit but wants to bug you and 'show off his script kiddie 3l337 magic tricks' :). This is easily done with a perl / python or bash shell endless loop inside which a random file named millions of empty directories instead of files is created.

To look up for empty directories hence use:

root@linux-server:~# DIR='/home';
root@linux-server:~# find  $DIR . -type d -empty > /root/empty_directories_list.txt

 

5. Delete all empty files only to clean up inodes

Deletion of empty files will automatically free up the inodes occupied, to delete them.

root@linux-server:~# cd /path/containing/multiple/empty-dirs/
root@linux-server:~# find . -type f -empty -exec rm -fr {} \;

 

6. Delete all empty directories only to clean up inocommanddes

root@linux-server:~# find . -type d -empty -exec rm -fr {} \;

 

7. Delete all empty files and directories to clean up inodes

root@linux-server:~# cd /path/containing/multiple/empty-dirs/
root@linux-server:~# find . -empty -delete

 

8. Use find + xargs to delete if files count to delete is too high

root@linux-server:~# find . -empty | xargs rm -r


That's all folks ! Enjoy now your Filesystem to have retrieved back the lost inodes from the jump empty files or directories.

Happy cleaning  🙂

How to list all files and directories in Windows XP drive and save it to a file

Friday, December 10th, 2010

Yesterday I had this thing, by accident because of some misconfiguration of the domain controller here in the university. I can read some files and folders which were belonging to students and teachers here in the university. We had some great fun with my good old friend David. In checking up some trashy stuff. It was rather interesting that one of the directory listings actually contained all the teachers computer usernames with it, therefore I needed a quick way to make all the files listing stored in a file in order to be able to later have this usernames listing stored in an MySQL or some other kind of database, after consulting few websites I came to the command to list all files and directories in the mapped sharing.

We first used the net use //workgroup/ J: to map the sharing in the workgroup and later we used the dir command as shown below to list all files and directories and store the data in a file.
So here is how you can also come with all your files and directories in your Windows drive stored in a file:

dir J: /s > c:somepathj_file.txt

Now if the files and directory is not too long you will have everything stored in j_file.txt in a couple of minutes time, however in our case the directory listing was huge so we had to wait for an hour and still the file was filling in 🙂

Howto delete empty directories in GNU /Linux with find linux command

Wednesday, April 7th, 2010

Ever wondered how you can delete all the empty directories in Linux?
I bet you did, there are many ways to achieve that in GNU/Linux, however here is one way you might go:
First it is probably a good idea to list the empty directories and examinethe empty directories before you take the next step and execute a command to delete them:

find . -depth -type d -empty

Now after you take a close look in the directories, next step to partake is delete the directories.

find . -depth -type d empty -exec rmdir {} ;

Be aware that in the above examples, the first one would list all directories in your current directory in which you
execute the command, the second example will delete all the empty directories starting from your current directory unto thedeepest located empty directory in the directory tree.