Hello Folks,
I've been playing around with FreeBSD for about 1 year now and i like it. Recently I've been reinstalling most of the Linux systems i use to use FreeBSD instead. Most of the systems are under minimal load and have lots of physical memory so i have not had any real issues. Last week I've decided to reinstall our MySQL server and before going live with it, I've run some tests of similar hardware and some test hardware. Here i started to run into massive memory using importing and exporting MySQL Schema. I'm testing with our largest DB, its about 5GB in size. Here is some test results:
1. Production Server Dell R610, 2 x X5675, 64GB RAM, 64GB SWAP, FreeBSD 12.1-p1, ZFS, MySQL-5.7.2.9, innodb_buffer_pool_size=40GB
The import took 1 Hour and 35 Minutes and consumed 100% of RAM and about 50% of SWAP (Live Server with other schemas and traffic but the load is minimal and Free RAM before i start the import was around 40 GB)
2. Test Server Dell T610 ESXi, FreeBSD 12.1-p1 VM, 1 CPU, 2GB RAM, 4 GB SWAP, ZFS, MySQL-5.7.2.9, innodb_buffer_pool_size=1GB/default config
Within 2 Minutes it consumed 100% of RAM, SWAP and died
3. Test Server Dell T610 ESXi, Debian 10.1.0 VM, 1 CPU, 2GB RAM, 2 GB SWAP, EXT4, MySQL-5.7.2.9, innodb_buffer_pool_size=128M/default config
Completed in 23 Minutes, RAM was in use about 95%, 0% of SWAP was used.
4.Test Server Dell T610 ESXi, Fedora 31 VM, 1 CPU, 2GB RAM, 2 GB SWAP, XFS, MySQL-5.7.2.9, innodb_buffer_pool_size=128M/default config
Completed in 21 Minutes, RAM was in use about 95%, 0% of SWAP was used.
5. Test Server Dell T610 ESXi, FreeBSD 12.1-p1 VM, 1 CPU, 2GB RAM, 4 GB SWAP, ZFS, MySQL-5.7.2.9, reduced innodb_buffer_pool_size=128M the rest is default
Within 2 Minutes it consumed 100% of RAM, SWAP and died. As if the innodb_buffer_pool_size value didnt reflect anything.
6. Test Desktop i7-8700 ESXi, FreeBSD 12.1-p1 VM, 1 CPU, 2GB RAM, 4 GB SWAP, ZFS, MySQL-5.7.2.9, innodb_buffer_pool_size=1GB/default config
Completed in 17 Minutes, 100% RAM was used and about 20% of the SWAP
As you can see in the test results. The Linux systems seem to do just fine where as BSD wants enormous amounts or it dies on the Dell Servers. Interesting the results are quiet different on a Desktop Computer. I have also seen this behavior on some of the other test VM systems with lesser memory when ClamAV performs system scan. All of the systems above are minimalist installs of the OS with only PKG, MySQL, SUDO, NANO, (OSSEC and CLAMAV on Production) and OpenVM Tools of used on ESXi.
Most of my rc.conf looks like this plus minus the application the server is used for:
Is there something I'm doing wrong? I've spent so many hours testing and reading manuals i think i'm going in circles now. Any help with greatly appreciate.
I've been playing around with FreeBSD for about 1 year now and i like it. Recently I've been reinstalling most of the Linux systems i use to use FreeBSD instead. Most of the systems are under minimal load and have lots of physical memory so i have not had any real issues. Last week I've decided to reinstall our MySQL server and before going live with it, I've run some tests of similar hardware and some test hardware. Here i started to run into massive memory using importing and exporting MySQL Schema. I'm testing with our largest DB, its about 5GB in size. Here is some test results:
1. Production Server Dell R610, 2 x X5675, 64GB RAM, 64GB SWAP, FreeBSD 12.1-p1, ZFS, MySQL-5.7.2.9, innodb_buffer_pool_size=40GB
The import took 1 Hour and 35 Minutes and consumed 100% of RAM and about 50% of SWAP (Live Server with other schemas and traffic but the load is minimal and Free RAM before i start the import was around 40 GB)
2. Test Server Dell T610 ESXi, FreeBSD 12.1-p1 VM, 1 CPU, 2GB RAM, 4 GB SWAP, ZFS, MySQL-5.7.2.9, innodb_buffer_pool_size=1GB/default config
Within 2 Minutes it consumed 100% of RAM, SWAP and died
3. Test Server Dell T610 ESXi, Debian 10.1.0 VM, 1 CPU, 2GB RAM, 2 GB SWAP, EXT4, MySQL-5.7.2.9, innodb_buffer_pool_size=128M/default config
Completed in 23 Minutes, RAM was in use about 95%, 0% of SWAP was used.
4.Test Server Dell T610 ESXi, Fedora 31 VM, 1 CPU, 2GB RAM, 2 GB SWAP, XFS, MySQL-5.7.2.9, innodb_buffer_pool_size=128M/default config
Completed in 21 Minutes, RAM was in use about 95%, 0% of SWAP was used.
5. Test Server Dell T610 ESXi, FreeBSD 12.1-p1 VM, 1 CPU, 2GB RAM, 4 GB SWAP, ZFS, MySQL-5.7.2.9, reduced innodb_buffer_pool_size=128M the rest is default
Within 2 Minutes it consumed 100% of RAM, SWAP and died. As if the innodb_buffer_pool_size value didnt reflect anything.
6. Test Desktop i7-8700 ESXi, FreeBSD 12.1-p1 VM, 1 CPU, 2GB RAM, 4 GB SWAP, ZFS, MySQL-5.7.2.9, innodb_buffer_pool_size=1GB/default config
Completed in 17 Minutes, 100% RAM was used and about 20% of the SWAP
As you can see in the test results. The Linux systems seem to do just fine where as BSD wants enormous amounts or it dies on the Dell Servers. Interesting the results are quiet different on a Desktop Computer. I have also seen this behavior on some of the other test VM systems with lesser memory when ClamAV performs system scan. All of the systems above are minimalist installs of the OS with only PKG, MySQL, SUDO, NANO, (OSSEC and CLAMAV on Production) and OpenVM Tools of used on ESXi.
Most of my rc.conf looks like this plus minus the application the server is used for:
Code:
hostname="test"
clear_tmp_enable="YES"
dumpdev="NO"
ifconfig_vmx0="inet 10.10.10.100 netmask 255.255.255.0"
ip6addrctl_enable="NO"
defaultrouter="10.10.10.10"
sshd_enable="YES"
ntpd_enable="YES"
ntpd_flags="-4"
zfs_enable="YES"
firewall_enable="YES"
firewall_script="/usr/local/etc/ipfw.rules"
firewall_logging="YES"
vmware_guest_vmblock_enable="YES"
vmware_guest_vmhgfs_enable="NO"
vmware_guest_vmmemctl_enable="YES"
vmware_guest_vmxnet_enable="YES"
vmware_guestd_enable="YES"
mysql_enable="yes"
ossec_hids_enable="YES"
clamav_clamd_enable="yes"
clamav_freshclam_enable="yes"
syslogd_enable="YES"
syslogd_flags="-4 -n -s -vv"
update_motd="NO"