Latest revision as of 15:38, 27 September 2018

Note: This page is old. Refer to this article instead.

Overview

This page describes the current wireless (WLAN) infrastructure at uniPHARM. As of September 2016, there is a business case to redo the entire infrastructure, so the info on this page is slated to change drastically.

There are 3 sets of access points in the current infrastructure: a set of 4 APs for the RF guns, a set of 2 APs for the mobile belt printers used in receiving, and a set of 2 APs for a guest wireless network connected to the Loomis network (DHL/Loomis). All SSIDs operate on the 2.4GHz; there are no 5GHz networks enabled as there is no current need for such.

Configuration

Access points

Access Point Group and Model	Hostname	Location	IP address	Wireless channel	Radio capability, current setting	Wireless MAC address	Admin login
RF Guns - Psion Teklogix 9160 G2
	ap7recv.unipharm.local	Receiving overhang	172.30.18.117	11	b/g, g-only	00:08:A2:08:38:90	admin/visionAP7
	ap13ship.unipharm.local	Shipping overhang	172.30.18.125	Disabled		00:08:A2:04:A7:94	admin/visionAP13
	ap9cage.unipharm.local	Forklift runway/Large Down	172.30.18.119	6		00:08:A2:08:3C:E0	admin/visionAP9
	ap12narc.unipharm.local	Narc cage	172.30.18.124	11		00:08:A2:04:C4:C8	admin/visionAP12
Belt Printer Access Points - Cisco AP541N-A-K9
	ap10c541n.unipharm.local	Receiving overhang	172.30.18.122	11	a/b/g/n, b/g/n	00:21:29:07:76:50	cisco/vision10
	ap11c541n.unipharm.local	Server room	172.30.18.123	11	a/b/g/n, b/g/n	64:AE:0C:EA:7C:00	cisco/vision11
Loomis Access Points - Cisco WAP4410N
	ap1guest	Narc cage	192.168.1.253	Disabled	b/g/n, n-only	adminit/vision1t
	ap2guest	Promo area (receiving)	192.168.1.252	Disabled	b/g/n, n-only	adminit/vision1t

Wireless networks

SSID	Purpose	AP Group	Security Type	Password	Notes
unitek2	RF gun network	Psion Teklogix 9160 G2	WPA-Enterprise	User: gunuser2, Password: keepscan9	Authentication server is internal RADIUS server within cluster.
uwddc1	Mobile belt printers	Cisco AP541N-K9	WEP	UWDVan20dyke52	Implemented because mobile belt printers had trouble connecting to Psion access points. (I've validated this - the printers don't handle WPA-Enterprise well. -norwizzle (talk))
unioffice2	Guest wireless (Walled garden)	Cisco WAP4410N	WPA2-Personal	Unipharm6y0d	Entire network disabled due to interference. Re-enabled upon visits from authorized guests.

Notes

02/17 - Access point failure causing massive and random outages

In Jan. to Feb. 2017, major issues were being experienced with the network serving the RF guns (Psion cluster). The guns would experience red locks (Lock-B) in TekTerm, which strongly indicates a loss of wireless connectivity (as opposed to the yellow Lock-H, which indicates being in host-wait i.e. attempting to resume communication with Bart). The most heavily affected area was OTC. After weeks of monitoring and investigation, it seems that one or more of the access points in the cluster are experiencing internal failure (likely the internal flash), due to age. Syslog indicated that some of the internal processes of the APs were restarting at random points in the day.

It is theorized that a failure of even one AP in the cluster has major ramifications to the operation of the entire cluster. This could be because wireless clients unknowingly roam/reassociate to the failing AP, or because any one access point contributes some processing in the cluster mechanism.

In an attempt to fix this, the AP8SHIP access point was replaced with AP13SHIP, a spare AP of the same model. During the replacement, the antenna connectors broke, such that the new AP did not seem to emit a strong enough signal for any of the guns to operate, despite being an active member (in other respects than wireless) of the cluster. Meanwhile, it was always a theory that the roaming between AP8SHIP and AP9CAGE to serve the main picking area (OTC,HTC,RX etc) was more detrimental than beneficial. This is due to the fact that the cluster is secured with WPA2-Enterprise - which requires slow, RADIUS authentication - coupled with the fact that no fast-roaming technologies are implemented (due to the client-side and AP-related barriers). Therefore, AP13SHIP is in place and running but not serving wireless (no wireless clients can connect to it).

Outcome: The Psion cluster serving the RF network has been effectively reduced to 3 APs (AP9cage,receiving, and narc). Feedback has been very positive thus far.

Misc. changes

MAC filtering turned off for Psion cluster. It's a useless and antiquated security measure.

To-do

Roaming and roaming authentication
Roaming stickiness
Transmit power
Presence of wireless b clients

@@ Line 1: / Line 1: @@
+ Note: This page is old. [[Information_Systems:Wireless_Infrastructure_(WLAN)|Refer to this article instead.]]
 =Overview=
 This page describes the current wireless (WLAN) infrastructure at uniPHARM. As of September 2016, there is a business case to redo the entire infrastructure, so the info on this page is slated to change drastically.
@@ Line 5: / Line 6: @@
 =Configuration=
-==Configuration Summary==
 ===Access points===
 {| border="1" class="wikitable" style="width: 90%"
@@ Line 21: / Line 21: @@
 | ap7recv.unipharm.local || Receiving overhang || 172.30.18.117 || 11 || rowspan=4 | b/g, g-only || 00:08:A2:08:38:90 || admin/visionAP7
 |-
-| ap13ship.unipharm.local || Shipping overhang || 172.30.18.125 || - ||  00:08:A2:04:A7:94  || admin/visionAP13
+| ap13ship.unipharm.local || Shipping overhang || 172.30.18.125 || ''Disabled'' ||  00:08:A2:04:A7:94  || admin/visionAP13
 |-
 | ap9cage.unipharm.local || Forklift runway/Large Down || 172.30.18.119 || 6 || 00:08:A2:08:3C:E0 || admin/visionAP9
@@ Line 35: / Line 35: @@
 | rowspan=3 | Loomis Access Points - Cisco WAP4410N
 |-
-| ap1guest || Narc cage || 192.168.1.253 || Disabled || rowspan=2 | b/g/n, n-only || adminit/vision1t
+| ap1guest || Narc cage || 192.168.1.253 || ''Disabled'' || rowspan=2 | b/g/n, n-only || adminit/vision1t
 |-
-| ap2guest || Promo area (receiving) || 192.168.1.252 || Disabled || adminit/vision1t
+| ap2guest || Promo area (receiving) || 192.168.1.252 || ''Disabled'' || adminit/vision1t
 |-
 |}
@@ Line 54: / Line 54: @@
 | uwddc1 || Mobile belt printers || Cisco AP541N-K9 || WEP || UWDVan20dyke52 || Implemented because mobile belt printers had trouble connecting to Psion access points. (I've validated this - the printers don't handle WPA-Enterprise well. -[[User:Norwinu|norwizzle]] ([[User talk:Norwinu|talk]]))
 |-
-| unioffice2 || Guest wireless (Walled garden) || Cisco WAP4410N || WPA2-Personal || Unipharm6y0d || '''Disabled due to interference'''
+| unioffice2 || Guest wireless (Walled garden) || Cisco WAP4410N || WPA2-Personal || Unipharm6y0d || '''Entire network disabled due to interference. Re-enabled upon visits from authorized guests.'''
 |}
-==Configuration Detail==
 =Notes=
+==02/17 - Access point failure causing massive and random outages==
+In Jan. to Feb. 2017, major issues were being experienced with the network serving the RF guns (Psion cluster). The guns would experience red locks (Lock-B) in TekTerm, which strongly indicates a loss of wireless connectivity (as opposed to the yellow Lock-H, which indicates being in host-wait i.e. attempting to resume communication with Bart). The most heavily affected area was OTC. After weeks of monitoring and investigation, it seems that one or more of the access points in the cluster are experiencing internal failure (likely the internal flash), due to age. Syslog indicated that some of the internal processes of the APs were restarting at random points in the day.
+It is theorized that a failure of even one AP in the cluster has major ramifications to the operation of the entire cluster. This could be because wireless clients unknowingly roam/reassociate to the failing AP, or because any one access point contributes some processing in the cluster mechanism.
-==Issues Log==
-* Errors related to internal memory/flash memory have been observed on AP7 (Receiving). Reboots are performed as a fix.
-* Severe connectivity issues with AP8 and AP9 caused a picking outage on Jan. 26, 2017. Upon further investigation, it seems the RADIUS component of the cluster glitched, causing the APs to behave erratically, connecting and disconnecting users. A staggered reboot of all Psion access points seems to have taken care of the problem.
-* AP12 (Narc cage) has an older firmware than the other 3 APs in the cluster. This should probably be upgraded, but meh. -[[User:Norwinu|norwizzle]] ([[User talk:Norwinu|talk]]) 16:27, 27 January 2017 (PST)
-[[Category:Networking]]
+In an attempt to fix this, the AP8SHIP access point was replaced with AP13SHIP, a spare AP of the same model. During the replacement, the antenna connectors broke, such that the new AP did not seem to emit a strong enough signal for any of the guns to operate, despite being an active member (in other respects than wireless) of the cluster. Meanwhile, it was always a theory that the roaming between AP8SHIP and AP9CAGE to serve the main picking area (OTC,HTC,RX etc) was more detrimental than beneficial. This is due to the fact that the cluster is secured with WPA2-Enterprise - which requires slow, RADIUS authentication - coupled with the fact that no fast-roaming technologies are implemented (due to the client-side and AP-related barriers). ''Therefore, AP13SHIP is in place and running but not serving wireless (no wireless clients can connect to it)''.
-==To-do==
+'''Outcome''': The Psion cluster serving the RF network has been effectively reduced to 3 APs (AP9cage,receiving, and narc). Feedback has been very positive thus far.
+==Misc. changes==
+* MAC filtering turned off for Psion cluster. It's a useless and antiquated security measure.
+=To-do=
 * Roaming and roaming authentication
 * Roaming stickiness
 * Transmit power
 * Presence of wireless b clients
+[[Category: Archived Pages]]

Difference between revisions of "Information Systems:Old Wireless (WiFi) Infrastructure (pre-2017)"