Some Notes on the Machines

SFTP, also called SCP, is not, accurately speaking, FTP. SFTP is “SSH File Transfer”, transfer of files over the SSH protocol, and not the FTP protocol. The FTP protocol has its own excellent secure (FTPS) capabilities, FileZilla being a recommended graphical FTP server for Windows; but GUI servers for SSH are much harder to find. Happily, here is one:

https://www.solarwinds.com/free-tools/free-sftp-server/registration?program=961&c=70150000000FJX0&CMP=BIZ-TAD-PCWDLD-SW_WW_X_PP_PPD_FT_EN_TOPFT_SW-SFTP-X_X_X_X-X

Here are two very helpful sources. For many general things, the EPEL repository:

http://fedoraproject.org/wiki/EPEL

and for PHP and other server-specifics, current versions and others more recent than the OSes supply by themselves, Remi:

http://rpms.remirepo.net/

An elegant solution from here:

https://social.technet.microsoft.com/Forums/systemcenter/en-US/e266d1eb-81ca-4084-8b3b-70f7c6904047/problem-with-inetresadmx?forum=winserverGP

Just go here:

https://www.microsoft.com/en-us/download/details.aspx?id=6243

and download the current versions of all of the templates (the above is for 2008R2, here’s to hoping Microsoft makes them available for all server versions), and replace them.

The indefatigable Matt Quick provides thus:

http://prajwaldesai.com/switching-between-gui-and-server-core-in-windows-server-2012/

Some great new guidance:

https://msdn.microsoft.com/en-us/library/windows/hardware/dn567661(v=vs.85).aspx

The automatic crawl scheduler in Windows Search Server sometimes fails; you can get far more positive and creative control with scripting. Here’s how.

First you’ll need the name of the “Search Service Application”. I found these under “Farm Search Administration” in the SharePoint site; the default/built-in is named “Search Service Application”, but on the server being worked there was a second custom one for production work, which for this document we will call “SSAXYZ”.

Now we need the “Content Sources” list. Here’s where we learn whether it’s working or not.

Next step is to create the scripts below. You may want all of them. A usually working location and filename is in each initial comment. If your server doesn’t have an E:, you will need to rewrite just a tad.

The way it works is, you set your scheduled task to run the CMD, and the CMD starts up the PowerShell. It is possible to have the scheduled task run the PowerShell directly, but it’s much harder to read that way. There are two pairs of scripts below, one pair for incremental crawls, and one for full. It is often recommendable to run the incremental daily and the full weekly, if available server horsepower (most of all RAID) is enough for the fileset(s) involved.

The PowerShell comes from here, it stops any crawling which may be going on (or stalled) and restarts the desired (full or incremental), for every Search Service Application. So if you have multiple SSA’s you may want to modify so that it only does what you want.

REM E:\Search Server Scripts\Initiate-Full-Crawl.cmd
C:\Windows\System32\WindowsPowerShell\v1.0\PowerShell.exe  " & ' E:\Search Server Scripts\\Initiate-Full-Crawl.ps1 ' "

# E:\Search Server Scripts\Initiate-Full-Crawl.ps1
Add-PSSnapin Microsoft.SharePoint.PowerShell -erroraction SilentlyContinue
Get-SPEnterpriseSearchCrawlContentSource -SearchApplication "SSAXYZ" | ForEach-Object {
    if ($_.CrawlStatus -ne "Idle")
    {
        Write-Host "Stopping currently running crawl for content source $($_.Name)..."
        $_.StopCrawl()
        do { Start-Sleep -Seconds 1 }
        while ($_.CrawlStatus -ne "Idle")
    }
    Write-Host "Starting full crawl for content source $($_.Name)..."
    $_.StartFullCrawl()
}

REM E:\Search Server Scripts\Initiate-Incremental-Crawl.cmd
C:\Windows\System32\WindowsPowerShell\v1.0\PowerShell.exe  " & ' E:\Search Server Scripts\\Initiate-Incremental-Crawl.ps1 ' "

# E:\Search Server Scripts\Initiate-Incremental-Crawl.ps1
Add-PSSnapin Microsoft.SharePoint.PowerShell -erroraction SilentlyContinue
Get-SPEnterpriseSearchCrawlContentSource -SearchApplication "SSAXYZ" | ForEach-Object {
    if ($_.CrawlStatus -ne "Idle")
    {
        Write-Host "Stopping currently running crawl for content source $($_.Name)..."
        $_.StopCrawl()        
        do { Start-Sleep -Seconds 1 }
        while ($_.CrawlStatus -ne "Idle")
    }
    Write-Host "Starting full crawl for content source $($_.Name)..."
    $_.StartIncrementalCrawl()
}

Completion of these steps will free up a lot of resources on an SBS. You will need to have a different method (e.g., Labtech, many others) to keep client machines up to date.

1. Run (install first if not present) the SQL Server Management Studio.
2. Connect the Studio, to type Database Engine, server name \\.\pipe\mssql$microsoft##ssee\sql\query
3. In the Studio, under Databases, you will find an item “SUSDB”.
4. If you don’t know where the database files are held, find them using the properties of that item.
5. Run an SQL query to get a list of connections: EXEC sp_who2
6. Run SQL query/ies to kill all connections to DBName SUSDB only, using the SPIDs, e.g., KILL 999
7. Right-click SUSDB, “Delete” SUSDB, and close the Studio.
8. Delete the database files and all of the update files.
9. Disable the WSUS-related items in services.msc and IIS.

The above is digested from the following:

https://social.technet.microsoft.com/Forums/en-US/d7f5d5bb-6623-4d46-80e6-421674a46829/remove-wsus-from-sbs2011-server?forum=smallbusinessserver

http://blog.mpecsinc.ca/2011/07/sbs-2011-wsus-sql-memory-usage-is-very.html

If it’s not working or not working well for you, try this:

1. In the Dashboard, Disable Office 365 integration and Azure integration. It’s two separate steps. If you disable just Office 365 integration you won’t be able to continue.
2. Set up Office 365 integration.
3. See lots of interesting information come up quite quicly, in the Office 365 tab after you restart the Dashboard, to confirm that your integration is working very nicely.
4. If it didn’t work well, open your firewall per Microsoft, and test again.

The great open-source tool Syncthing works very well for private DropBox-class capability, including optional file-multiversioning at one or more sync locations. I have some very used PC hardware which would work well enough for this, but then I was thinking, I could save the cost of a 24/7 150W light bulb at the very least, if my cute little Raspberry Pi 2 could do the job.

The end of the story is already here; it’s working beautifully. Here’s the elements I am using:

• One Raspberry Pi 2 with Raspbian installed in default fashion on its bundled microSD card, good passwords set, and fully up to date, including firmware. I have been finding considerable performance boosts with update of firmware, so I definitely recommend it.
• One USB flash drive big enough to hold your sync archive with satisfactory free space. It might work with a USB hard drive, but it didn’t work with my OEM-portablized Hitachi TravelStar, too much power demand…although I did limit the power use of the whole rig by using one 2.1A double-USB cell-phone charger for everything! :-) I’m now using a 32-gigabyte LEXAR USB flash drive, the archive is currently 17.5G.
• Wire or wireless to the Internet connection. I have not seen a major performance hit from wireless, but then again I’m using a pretty beefy wireless router and my wifi environment is not choked.
• The USB flash drive formatted BTRFS. (One does the format using mkfs.btrfs, and then sets the /etc/fstab for automount and auto-check at boot.) This is important, because one can find quite a large number of reports of USB flash drives failing when used as OS drives under ext3, ext4, and NTFS. I first set it all up using NTFS on the flash drive, and had noticed the flash drive’s LED flashing frankly all of the time, even in moments when there was no traffic according to Syncthing, and the incoming stream was forced to pause a lot. That concerned me, so I tried it in ext4, and it was just about as bad. So I did a bit of research, found the many reports of failures, found multiple flash-specific filesystems all of which admitted of various different downsides which I didn’t like very much, set it up again using BTRFS, and was really amazed. The LED went off a whole lot less, and the effective incoming bandwidth during full sync literally doubled, ~1-1.5M/s to ~3-3.8M/s depending on the moment, and no pauses visible.
• Special BTRFS attributes set up in /etc/fstab. I set these and did a reboot after the full sync was about one-third done, and was yet more amazed. Even when full sync was going on, the LED did not and does not flash more than a half-second or so every once in a long while, and incoming stream bandwidth remains very nice. I’m using ssd_spread, noatime, and space_cache. A USB stick can’t really count as an SSD because its circuitry hides a lot of things from the OS, so the attribute “ssd” does not take effect if set, but these three work just fine and should keep it running as long and reliably as possible.

Here it is on my bench at home, doing its so-quiet production work, not moved yet to its permanent location. Everything except the (temporary) screen is visible, the “power supply” is plugged into a two-prong extension cord; and those are CD blanks in the corner for a size comparison :-)

Try it, you’ll like it! :-)

More methods of the same, not tested yet:

https://www.seafile.com

Lots of devices and old PCs have trouble connecting to Server 2012, because Server 2012 uses SMB 3.0, and older devices often use SMB 1.0. There is a way:

http://xenomorph.net/windows/issues/windows-xp-cannot-connect-to-2012-r2-file-shares/